Association of urinary metal exposure with H. pylori seropositivity and mortality in the U.S. Population: NHANES 1999–2000 analysis

Abstract

Background

The relationship between urinary heavy metal concentrations and overall mortality remains understudied, particularly among individuals with Helicobacter pylori (H. pylori) infection. This study aimed to explore the associations between urinary concentrations of specific metals and H. pylori seropositivity, as well as to evaluate their respective associations with all-cause mortality in both H. pylori-infected and non-infected subpopulations.

Methods

Data were derived from the 1999–2000 cycles of the National Health and Nutrition Examination Survey (NHANES), resulting a final analytical cohort of 1,086 participants. Multivariable logistic regression models were applied to identify predictors of H. pylori infection. Cox proportional hazards models and restricted cubic spline analyses were used to investigate associations between urinary metal concentrations and all-cause mortality, stratified by H. pylori status.

Results

Restricted cubic spline analyses revealed a significant association between urinary thallium levels and all-cause mortality in H. pylori-positive individuals (P for nonlinearity = 0.2731), with a nonlinear relationship observed in H. pylori-negative participants (P for nonlinearity = 0.0127). Elevated urinary tungsten concentrations were independently associated with H. pylori seropositivity (P overall = 0.0074). Notably, among H. pylori-negative participants, the highest tungsten quartile (Q4) showed a significant mortality risk increase (HR = 3.054, 95% CI: 1.083–8.608), while in H. pylori-positive individuals, tungsten was significantly associated with mortality only in continuous models (HR = 3.129, 95% CI: 1.031–9.493).

Conclusion

Elevated urinary tungsten levels were significantly linked to H. pylori infection, while the relationship between tungsten and mortality differed by infection status: a dose–response association was observed in H. pylori-negative individuals, while a nonlinear association emerged in H. pylori-positive participants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-025-04223-0.

Introduction

Helicobacter pylori (H. pylori) infection, a prevalent colonization of the gastric mucosa, infects over 4.4 billion people worldwide and remains a significant etiological factor in various gastrointestinal disorders [1, 2]. Extensive research has focused on elucidating the mechanisms by which H. pylori contributes to oxidative stress, marked by an imbalance between reactive oxygen species (ROS) generation and antioxidative defenses [3]. The activation of inflammatory pathways and disruptions in cellular redox balance are key factors in the development of H. pylori-related conditions such as gastritis, peptic ulcers, and gastric cancer [4].

Recent meta-analyses have reported altered serum concentrations of several metals, including zinc, selenium, and manganese, in individuals infected with H. pylori [5–7]. However, these studies have primarily focused on metal concentrations in serum and lacked long-term outcome data, such as mortality. In addition, a meta-analysis of 21 original studies conducted by Shen demonstrated a significant increase in serum concentrations of chromium, nickel, and mercury in leukemia patients, accompanied by a noteworthy decrease in serum manganese levels [8]. Additionally, another study by Rahman showed that contact with metals like Thallium and Tungsten would increase the risks of stroke [9]. Epidemiologic evidence underscores the increasing influence of metals on human health with industrial progression, revealing their detrimental relationships with human health.

Prior investigations have indicated a potential correlation between heavy metal exposure and an elevated susceptibility to Helicobacter pylori (H. pylori) infection. Research findings demonstrate variations in the concentration of serum elements, including zinc, cobalt, and magnesium [5], among individuals infected with H. pylori compared to those uninfected [5, 10]. Notably, the study by Krueger and Wade [11] serves as a pivotal precedent in this research domain, identifying significant associations between blood levels of lead and cadmium and H. pylori seropositivity. Our study builds upon this foundational work by exploring a broader range of urinary metals and incorporating long-term mortality outcomes. This association suggests that the immunosuppressive associations attributed to lead and cadmium toxicity might contribute to an augmented susceptibility to H. pylori infection. These insights underscore the intricate interplay between heavy metal exposure and the dynamics of H. pylori infection, providing a foundation for further exploration into the mechanistic underpinnings of this relationship. However, most current research is predominantly focused on a limited selection of metals, and are often constrained in the insufficient mortality analysis. Tungsten, a dense transition metal known for its high melting point and strength, has been increasingly utilized in industrial, military, and consumer applications. Historically, Tungsten exposure has been associated with mining and smelting activities; however, more recently, environmental sources such as contaminated groundwater, industrial emissions, and Tungsten-based products have contributed to widespread human exposure. Despite its widespread use, the health effects of Tungsten remain inadequately understood. Emerging evidence suggests that Tungsten may exert toxicological effects, including carcinogenicity, immune dysfunction, and reproductive toxicity [12, 13]. Given the potential for environmental and occupational exposure, it is essential to elucidate the health risks associated with Tungsten.

Therefore, this study investigates the relationship between urinary metal levels and H. pylori infection, and to further explore the relationship between urinary metal exposure and all-cause mortality in individuals stratified by H. pylori infection status using data from the NHANES database.

Materials and methods

Study design and participants

The National Health and Nutrition Examination Survey (NHANES) utilizes a rigorous, multistage probability sampling design to recruit a nationally representative cohort of the U.S. population. Its core objective is to assess the health and nutritional status of Americans [14]. The survey protocol is approved by the Institutional Review Board of the National Center for Health Statistics, and written informed consent is obtained from all participants. NHANES gathers a wide array of data, including demographic information, clinical measurements, dietary intake, laboratory test results, and responses to standardized questionnaires [15].

During the 1999–2000 NHANES cycles, a total of 8,344 participants were initially enrolled. After excluding participants lacking data on H. pylori infection status or urinary metal measurements, as well as participants lost to follow-up or not meeting infection criteria, the final analytic sample was established (Fig. 1).

Fig. 1.

Flowchart of participant selection from the NHANES 1999–2000. This flowchart illustrates the inclusion and exclusion criteria applied to derive the final analytical cohort (n = 1,086), based on availability of urinary metal measurements, H. pylori serostatus, and mortality follow-up information

Urinary metal measurements

Between 1999 and 2000, the NHANES urinary metal analysis consistently included measurements of ten metal elements in each survey cycle. These elements comprised Barium, Beryllium, Cadmium, Cobalt, Cesium, Molybdenum, Lead, Platinum, Antimony, Thallium, and Tungsten. Random (spot) urine samples were collected from participants following confirmation of the absence of background contamination in the collection materials. Urinary creatinine concentrations were measured and included as covariates in all multivariable regression models (Tables 1, 2, 3, 4, 5 and 6) to adjust for urinary dilution. We opted for the covariate-adjustment approach rather than using creatinine-standardized metal concentrations (e.g., ng/g creatinine), as this method minimizes bias from creatinine-associated physiological factors and better preserves the continuous distribution of metal exposures for spline modeling and interpretation. This approach is consistent with current practices in NHANES-based biomonitoring studies. While spot urine samples are commonly used in biomonitoring studies, they may introduce exposure misclassification due to intra-individual variability and short biological half-lives. To ensure the accuracy of metal concentration measurements, samples underwent rigorous processing, storage, and transportation protocols before analysis. The analysis itself involved direct measurement using mass spectrometry, a highly sensitive technique capable of providing precise quantification of metal concentrations in urine samples. This standardized approach across survey cycles establishes a robust foundation for evaluating the trends and associations related to the specified metal elements over the designated time frame. Specific details can be seen on the website (https://wwwn.cdc.gov/Nchs/Nhanes/1999-2000/LAB06HM.htm#WTSHM04).

Table 1.

Associations between the Urinary Lead and H. pylori infection

Urinary Lead	Model 0 1.238(1.053,1.456) *	Model 1a 1.106(0.895,1.367)	Model 2b 1.138(0.891,1.455)
Q1	ref	ref	ref
Q2	1.084(0.619,1.898)	0.919(0.430,1.964)	0.909(0.436,1.894)
Q3	1.260(0.676,2.347)	0.756(0.310,1.842)	0.828(0.368,1.862)
Q4	1.948(1.050,3.615) *	1.255(0.615,2.561)	1.304(0.564,3.014)

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, Urinary Thallium and Urinary Tungsten

^*P < 0.05

Table 2.

Associations between the Urinary Thallium and H. pylori infection

Urinary Thallium	Model 0 0.378(0.180,0.795) *	Model 1a 0.639(0.128,3.201)	Model 2b 0.308(0.040,2.368)
Q1	ref	ref	ref
Q2	1.032(0.563,1.890)	1.008(0.396,2.566)	0.877(0.436,1.762)
Q3	0.650(0.385,1.098)	0.577(0.328,1.012)	0.433(0.245,0.766) **
Q4	0.740(0.381,1.437)	0.912(0.390,2.131)	0.679(0.284,1.623)

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, urinary Lead and urinary Tungsten

^*P < 0.05

^**P < 0.01

Table 3.

Associations between the urinary tungsten and H. pylori infection

Urinary Tungsten	Model 0 1.958(1.214,3.158) *	Model 1a 2.950(1.369,6.359) *	Model 2b 3.003(1.339,6.734) *
Q1	ref	ref	ref
Q2	0.665(0.428,1.034)	0.736(0.359,1.506)	0.740(0.408,1.341)
Q3	0.986(0.567,1.715)	1.024(0.431,2.435)	1.054(0.468,2.376)
Q4	1.176(0.774,1.788)	1.557(0.950,2.551)	1.576(0.910,2.732)

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, urinary Lead and urinary Thallium

^*P < 0.05

Table 4.

Associations of the Urinary Lead and all-cause mortality in H. pylori positive and negative infected participants

Variables	Model 0	Model 1a	Model 2b
HR (95%CI) in H. pylori positive
Continuous variable	0.82(0.63,1.06)	0.77(0.58,1.02)	0.75(0.52,1.07)
Q1	ref	ref	ref
Q2	0.82(0.40,1.71)	0.85(0.30,2.39)	1.30(0.56,3.02)
Q3	0.69(0.24,1.98)	0.84(0.19,3.69)	2.06(0.84,5.05)
Q4	0.56(0.23,1.40)	0.64(0.19,2.17)	0.77(0.23,2.56)
HR (95%CI) in H. pylori negative
Continuous variable	1.233(1.052,1.446) *	1.045(0.842,1.297)	1.127(0.925, 1.374)
Q1	ref	ref	ref
Q2	1.32(0.57,3.09)	0.79(0.11,5.56)	1.45(0.32, 6.48)
Q3	1.50(0.82,2.76)	0.62(0.11,3.39)	1.06(0.27, 4.20)
Q4	1.67(0.91,3.08)	0.73(0.15,3.56)	1.10(0.32, 3.73)

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, urinary Thallium and urinary Tungsten

^*P < 0.05

Table 5.

Associations of the urinary thallium and all-cause mortality in H. pylori positive and negative infected participants

Variables	Model 0	Model 1a	Model 2b
HR (95%CI) in H. pylori positive
Continuous variable	0.011(0.001,0.088) ***	0.046(0.004,0.580) *	0.165(0.007,3.851)
Q1	ref	ref	ref
Q2	0.526(0.254,1.093)	0.640(0.214,1.911)	0.714(0.362, 1.408)
Q3	0.448(0.140,1.437)	0.636(0.177,2.284)	0.835(0.228, 3.059)
Q4	0.209(0.100,0.435) ***	0.302(0.121,0.752) *	0.471(0.161, 1.381)
HR (95%CI) in H. pylori negative
Continuous variable	0.108(0.020,0.580) *	0.130(0.004,4.096)	0.144(0.005, 4.332)
Q1	ref	ref	ref
Q2	1.140(0.574,2.263)	0.830(0.452,1.525)	0.712(0.434, 1.166)
Q3	0.536(0.292,0.986) *	0.378(0.103,1.383)	0.348(0.087, 1.390)
Q4	0.460(0.235,0.902) *	0.515(0.146,1.821)	0.537(0.180, 1.603)

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, urinary Lead and urinary Tungsten

^*P < 0.05

^***P < 0.001

Table 6.

Associations of the urinary tungsten and all-cause mortality in H. pylori positive and negative infected participants

Variables	Model 0	Model 1a	Model 2b
HR (95%CI) in H. pylori positive
Continuous variable	1.565(1.024,2.391) *	2.265(1.246,4.119) *	3.129(1.031,9.493) *
Q1	ref	ref	ref
Q2	0.752(0.334,1.695)	0.878(0.349,2.209)	0.903(0.405, 2.011)
Q3	0.619(0.341,1.121)	0.808(0.336,1.945)	1.193(0.460, 3.095)
Q4	0.665(0.327,1.350)	1.228(0.373,4.041)	1.566(0.239,10.270)
HR (95%CI) in H. pylori negative
Continuous variable	0.315(0.066,1.499)	0.970(0.137,6.884)	1.735(0.333, 9.031)
Q1	ref	ref	ref
Q2	1.352(0.623,2.935)	1.781(0.639,4.964)	3.052(1.211, 7.690) *
Q3	1.901(1.074,3.364) *	2.992(1.653,5.413) **	3.372(1.606, 7.080) **
Q4	0.853(0.517,1.409)	1.490(0.481,4.618)	3.054(1.083, 8.608) *

^aModel 1 adjusted for age, sex, races and education

^bModel 2 adjusted for age, sex, waist, poverty, races, education, HbA1c, alcohol drink, HDL, TG, diabetes mellitus, hypertension, urinary creatinine, urinary Lead and urinary Thallium

^*P < 0.05

^**P < 0.01

Helicobacter pylori status

H. pylori immunoglobulin G (IgG) antibodies were quantified using an enzyme-linked immunosorbent assay (ELISA) kit from Wampole Laboratories (Cranbury, NJ) [16]. Based on ELISA optical density (OD) thresholds, participants were classified as H. pylori positive (OD ≥ 1.1) or negative (OD < 0.9). Samples with equivocal OD values ranging from 0.9 to 1.1 were excluded to maintain analytical precision [17].

Follow up and endpoint

Mortality status and cause of death were determined via linkage to the National Death Index (NDI) database (https://www.cdc.gov/nchs/data-linkage/mortality.htm), with follow-up through December 31, 2019. Median follow-up was 234 months (IQR: 173–243) for H. pylori-positive individuals and 236 months (IQR: 229–243) for H. pylori-negative participants.

Covariate

Several covariates potentially confounding the association between urinary metal exposure and H. pylori infection were included: age, sex, races, poverty status, educational attainment (categorized as < high school, high school, or > high school), smoking status, alcohol use, diabetes mellitus, hypertension, cardiovascular disease, serum lipid profiles, and blood glucose levels. Hypertension was defined by any of: self-reported physician diagnosis, systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or current antihypertensive medication use [18]. Smoking status comprised: current (smoked ≥ 100 cigarettes lifetime and within past month), former (smoked ≥ 100 cigarettes but not current), and never smokers (< 100 cigarettes lifetime) [19]. Alcohol consumption categories were: never (lifetime abstainers), former (abstinent ≥ 1 year), moderate (≤ 1 drink/day [women] or ≤ 2 drinks/day [men]), and heavy (> moderate limits and/or frequent binge drinking) [20, 21]. Additionally, educational level was classified into three groups: less than high school, high school, and more than high school. Estimated glomerular filtration rate (eGFR) was calculated using the CKD-EPI equation [22]. Participants missing urinary metal data, H. pylori status, or follow-up information were excluded. To preserve sample size, those missing covariate data were retained in multivariable models if all variables for that specific model were available. Participants with missing data on primary exposures (i.e., urinary metals), H. pylori serostatus, or mortality were excluded from the analysis. For covariates with missing values, we performed single imputation using the median (for continuous variables) or mode (for categorical variables), depending on the variable type.

Statistical analysis

The NHANES utilizes a sophisticated multistage probability sampling framework that incorporates stratification, clustering, and weighting procedures. Statistical analyses within this framework rely primarily on the survey’s complex design rather than on assumptions about the data’s underlying distribution. Unlike traditional methods, the reliability of parametric tests in complex survey data does not require normality assumptions; instead, correct application of sampling weights is essential to yield unbiased estimates representative of the target population. The weighted Student’s t-test is commonly applied in such analyses, as it adjusts for design elements including sampling weights, stratification, and primary sampling units, thereby providing valid inference that generalizes to the U.S. civilian non-institutionalized population [23]. Baseline characteristics were compared between H. pylori-positive and H. pylori-negative participants using descriptive statistics. Continuous variables, assessed for normality via the Kolmogorov–Smirnov test, are presented as mean ± standard deviation (SD) and compared using Student's t-tests or Wilcoxon rank-sum tests, as appropriate. Categorical variables are summarized as frequencies (percentages) and compared using Chi-square tests.

Univariable and multivariable-adjusted logistic regression models estimated odds ratios (ORs) and 95% confidence intervals (CIs) for the association between urinary metal concentrations and H. pylori infection status. Additionally, restricted cubic spline (RCS) models were constructed to assess potential nonlinear relationships between urinary metal concentrations and mortality risk. Knots were placed at the 5th, 35th, 65th, and 95th percentiles of the urinary metal distribution and the selection of four knots follows established epidemiological modeling practices, balancing model flexibility with the risk of overfitting, and is consistent with prior studies using RCS for exposure–response analyses. Stratified analyses by H. pylori infection status were conducted to evaluate differential associations [24]. Examining the association of urinary metals with all-cause mortality, cox proportional hazards models were utilized to compute hazard ratios (HRs) and their corresponding 95% Cis [25]. Multivariable models were constructed by including covariates selected a priori based on biological plausibility, clinical relevance, and established literature. No automated stepwise procedures were applied, in order to reduce the risk of overfitting and improve model interpretability: Model 1 adjusted for age, sex, BMI, and poverty; Model 2 additionally controlled for race, education, smoking status, alcohol use, and waist circumference; Model 3 further incorporated HDL cholesterol, diabetes mellitus, hypertension, uric acid, eGFR, and urinary concentrations of Lead, Thallium, and Tungsten, alongside covariates in Model 2. Urinary metals were categorized into quartiles, with the lowest quartile (Q1) serving as the reference. Cox proportional hazards regression was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality, analyzing metals both continuously and by quartile, applying the same covariate adjustment strategy as in logistic models. Kaplan–Meier survival curves stratified by quartiles were generated and compared via log-rank tests (Figs. 2, 3 and 4). The proportional hazards assumption was not formally tested due to statistical limitations, which we acknowledge as a methodological limitation of our analysis. To strengthen the credibility of our results, we calculated the E-value to evaluate the impact of unmeasured and unknown confounders on the association [26]. We also made variance inflation factor (VIF) test to check the linearity assumption in logistic regression models. To control for multiple testing, p-values were adjusted using the Benjamini–Hochberg (BH) procedure to control the false discovery rate (FDR).

Fig. 2.

Kaplan–Meier survival curves for all-cause mortality by urinary Lead quartiles in (A) H. pylori-positive and (B) H. pylori-negative participants. Survival differences were assessed using log-rank tests. Unadjusted survival estimates are shown

Fig. 3.

Kaplan–Meier survival curves for all-cause mortality by urinary Thallium quartiles in (A) H. pylori-positive and (B) H. pylori-negative individuals. Unadjusted survival estimates are shown

Fig. 4.

Kaplan–Meier survival curves for all-cause mortality by urinary Tungsten quartiles in (A) H. pylori-positive and (B) H. pylori-negative participants. Unadjusted survival estimates are shown

A two-sided P < 0.05 was considered statistically significant. All analyses were performed using SPSS version 26.0 (IBM Corp, Armonk, NY) and R (version 4.3.2) [27, 28].

Results

Study participants and baseline characteristics

The final cohort included 1,086 American adults, among whom 468 participants tested positive for H. pylori (Fig. 1 and Table 7). The average age was 46.13 ± 19.95 years, with Males accounting for 47.4% of the sample. In the overall participant group, the mean urinary lead concentration was found to be 1.32 ± 1.54 ng/mL. In addition, the mean urinary Thallium concentration was 0.21 ± 0.14 ng/mL, urinary lead levels measured 1.32 ± 1.54 ng/mL, and the mean urinary Tungsten concentration was 0.14 ± 0.31 ng/mL. Specifically, participants with H. pylori infection had older age, greater waist circumference, lower socioeconomic and educational levels, current smoking, former alcohol coOur study explored nsumption, as well as increased rates of diabetes, hypertension, elevated BMI, and higher fasting triglyceride concentrations.

Table 7.

Baseline characteristics of participants with different H. pylori infection status

	Overall (n = 1086)	Hp negative (n = 618)	Hp positive (n = 468)	P value
Age (years)	46.13 (19.95)	43.41 (19.54)	49.71 (19.93)	< 0.001
Sex (male %)	515 (47.4)	290 (46.9)	225 (48.1)	0.753
BMI (kg/m2)	28.04 (6.34)	27.73 (6.22)	28.44 (6.47)	0.066
Waist (cm)	95.55 (15.96)	94.69 (15.90)	96.70 (16.00)	0.041
Race (%)				< 0.001
Mexican American	252 (23.2)	86 (13.9)	166 (35.5)
Non-Hispanic Black	234 (21.5)	107 (17.3)	127 (27.1)
Non-Hispanic White	495 (45.6)	372 (60.2)	123 (26.3)
Other Hispanic	69 (6.4)	29 (4.7)	40 (8.5)
Other Race	36 (3.3)	24 (3.9)	12 (2.6)
Poverty income ratio	2.55 (1.65)	2.84 (1.65)	2.15 (1.57)	< 0.001
Education (%)				< 0.001
High School	274 (25.3)	163 (26.5)	111 (23.8)
Less Than High School	393 (36.4)	138 (22.4)	255 (54.7)
More Than High School	414 (38.3)	314 (51.1)	100 (21.5)
Fasting insulin (uU/mL)	13.52 (12.70)	13.25 (12.49)	13.91 (13.01)	0.566
Fasting glucose (mg/dL)	103.46 (43.05)	100.47 (35.50)	107.88 (52.02)	0.056
HbA1c (%)	5.50 (1.11)	5.38 (0.98)	5.65 (1.24)	< 0.001
Total bilirubin (umol/L)	0.57 (0.28)	0.57 (0.28)	0.57 (0.27)	0.672
Urinary creatinine (mg/dL)	0.73 (0.44)	0.73 (0.39)	0.74 (0.50)	0.936
eGFR	97.27 (25.34)	98.26 (24.93)	95.97 (25.85)	0.143
Uric acid (mg/dL)	5.29 (1.53)	5.23 (1.48)	5.37 (1.60)	0.133
TC (mg/dL)	200.36 (42.42)	199.88 (42.02)	200.99 (42.98)	0.669
TG (mg/dL)	142.53 (99.57)	135.26 (86.99)	153.28 (115.11)	0.045
LDL (mg/dL)	124.68 (35.29)	126.31 (35.06)	122.23 (35.59)	0.234
HDL (mg/dL)	50.98 (14.39)	52.22 (14.58)	49.35 (13.99)	0.001
Smoke (%)				0.166
Former	261 (26.9)	148 (27.3)	113 (26.5)
Never	494 (51.0)	287 (52.9)	207 (48.6)
Current	214 (22.1)	108 (19.9)	106 (24.9)
Alcohol drink (%)				0.012
Former	204 (22.0)	101 (19.1)	103 (25.8)
Heavy	164 (17.7)	96 (18.1)	68 (17.0)
Mild	305 (32.9)	186 (35.2)	119 (29.8)
Moderate	128 (13.8)	83 (15.7)	45 (11.3)
Never	127 (13.7)	63 (11.9)	64 (16.0)
Hypertension (%)	393 (36.2)	188 (30.4)	205 (43.8)	< 0.001
SBP (mmHg)	125.00 (21.08)	122.21 (20.23)	128.76 (21.64)	< 0.001
DBP (mmHg)	71.01 (13.98)	70.40 (13.91)	71.83 (14.04)	0.1
Diabetes Mellitus (%)	112 (10.9)	50 (8.7)	62 (13.7)	0.03
CRP (mg/dL)	0.48 (0.80)	0.45 (0.76)	0.51 (0.84)	0.179
Stroke (%)	32 (3.3)	13 (2.4)	19 (4.4)	0.106
Urinary Barium (ng/mL)	2.34 (4.13)	2.55 (4.67)	2.06 (3.25)	0.05
Urinary Beryllium (ng/mL)	0.09 (0.00)	0.09 (0.00)	0.09 (0.00)	0.384
Urinary Cobalt (ng/mL)	0.58 (1.94)	0.61 (2.52)	0.54 (0.58)	0.594
Urinary Cesium (ng/mL)	5.49 (4.14)	5.47 (3.31)	5.52 (5.02)	0.874
Urinary Molybdenum (ng/mL)	62.49 (56.50)	63.16 (59.52)	61.61 (52.29)	0.654
Urinary Lead (ng/mL)	1.32 (1.54)	1.16 (1.34)	1.54 (1.76)	< 0.001
Urinary Platinum (ng/mL)	0.03 (0.06)	0.03 (0.08)	0.03 (0.00)	0.372
Urinary Antimony (ng/mL)	0.18 (0.24)	0.19 (0.29)	0.18 (0.17)	0.717
Urinary Thallium (ng/mL)	0.21 (0.14)	0.22 (0.13)	0.20 (0.14)	0.031
Urinary Tungsten (ng/mL)	0.14 (0.31)	0.13 (0.17)	0.17 (0.43)	0.024

Abbreviation: Hp Helicobacter Pylori, SBP systolic blood pressure, DBP diastolic blood pressure, BMI body mass index, HbA1c glycated hemoglobin, eGFR estimated glomerular filtration rate, TC total cholesterol, TG triglyceride, HDL high-density lipoprotein, LDL low-density lipoprotein, CRP C-reactive protein

Associations between urinary metal concentrations and H. pylori infection

To investigate factors associated with H. pylori infection, we first performed univariate logistic regression analyses using H. pylori infection status (positive vs. negative) as the binary outcome variable. Covariates that demonstrated statistical significance (P < 0.05) in univariate models, including age, sex, poverty status, and HDL, were considered for inclusion in multivariable models (Table 8).

Table 8.

Risk factors for H. pylori infection in adults in NHANES 1999–2018

Variables	β	Standard Error	P value	95% CI
Age	0.01	0.00	< 0.001	0.01 (0.01 ~ 0.01)
Sex	−0.26	0.04	< 0.001	−0.26 (−0.35 ~ −0.17)
Race	0.00	0.04	0.950	0.00 (−0.08 ~ 0.08)
Education	−0.15	0.11	0.150	−0.15 (−0.36 ~ 0.06)
BMI	0.01	0.01	0.074	0.01 (−0.00 ~ 0.03)
Waist	−0.00	0.00	0.093	−0.00 (−0.01 ~ 0.00)
Poverty	−0.04	0.01	< 0.001	−0.04 (−0.06 ~ −0.02)
Diabetes mellitus	−0.10	0.07	0.180	−0.10 (−0.24 ~ 0.05)
Hypertension	0.04	0.04	0.246	0.04 (−0.03 ~ 0.11)
Smoke	−0.05	0.04	0.153	−0.05 (−0.13 ~ 0.02)
Alcohol drink	0.21	0.12	0.071	0.21 (−0.02 ~ 0.44)
CRP	−0.00	0.02	0.986	−0.00 (−0.05 ~ 0.05)
Total bilirubin	−0.04	0.07	0.533	−0.04 (−0.17 ~ 0.09)
Urinary creatinine	0.05	0.07	0.473	0.05 (−0.09 ~ 0.19)
Uric acid	0.02	0.01	0.090	0.02 (−0.00 ~ 0.05)
eGFR	−0.00	0.05	0.995	−0.00 (−0.10 ~ 0.10)
Urinary Barium	−0.00	0.00	0.426	−0.00 (−0.01 ~ 0.01)
Urinary Beryllium	0.01	0.00	0.114	0.01 (−0.00 ~ 0.02)
Urinary Cobalt	0.02	0.04	0.725	0.02 (−0.07 ~ 0.10)
Urinary Cesium	−0.00	0.00	0.521	−0.00 (−0.00 ~ 0.00)
Urinary Molybdenum	−0.03	0.07	0.633	−0.03 (−0.16 ~ 0.10)
Urinary Lead	0.01	0.00	0.002	0.01 (0.01 ~ 0.01)
Urinary Antimony	−0.25	0.23	0.294	−0.25 (−0.71 ~ 0.21)
Urinary Platinum	−0.17	0.15	0.250	−0.17 (−0.47 ~ 0.12)
Urinary Thallium	0.14	0.05	0.014	0.14 (0.03 ~ 0.24)
Urinary Tungsten	0.02	0.02	0.257	0.02 (−0.01 ~ 0.05)

We then assessed the associations between urinary metal concentrations and H. pylori infection. Among the metals evaluated, urinary concentrations of lead, Thallium, and Tungsten were found to be significantly associated with H. pylori infection in either baseline comparisons (Table 7) or univariate regression analyses (Table 8), and were therefore selected for further analysis.

Multivariable logistic regression models were constructed to examine whether these metals remained independently associated with H. pylori infection after adjusting for potential confounders. When modeled as continuous variables, urinary Lead showed significant positive associations in the model 0, but these associations were attenuated and became nonsignificant after adjusting for covariates in subsequent models (model 1–2) (Tables 7). In Table 2, urinary Thallium was negatively associated with H. pylori infection in model 0, while the associations didn’t continue after adjustments for covariates in model 1–2. In contrast, urinary Tungsten maintained a significant positive association with H. pylori infection across all models (model 0 to model 2), suggesting a robust and independent relationship (Table 3). However, when urinary Tungsten was analyzed in quartiles, participants in the highest quartile (Q4) had higher odds of H. pylori infection compared to those in the lowest quartile (Q1) without significances (HR 1.576, 95% CI: 0.910–2.732), after full adjustment in model 2. These findings indicate that among the metals analyzed, Tungsten exhibited the most consistent and independent association with H. pylori seropositivity. Although the confidence interval is relatively wide, its exclusion of the null indicates a statistically and potentially clinically meaningful association. In supplementary Table 1, we Made VIF to evaluate the adjusted factors in model 2 and all confounding factors showed no multicollinearity. After FDR correction using the BH method, urinary tungsten is still significantly associated with H. pylori infection (P value after adjustment = 0.0375). The details were shown in supplementary Table 2.

Correlation between urinary metal concentrations and all-cause mortality, stratified by H. pylori status

To assess the association between urinary metal concentrations and all-cause mortality, we performed Cox proportional hazards regression analyses using follow-up data through December 31, 2019. Mortality risk was evaluated separately in H. pylori-positive and H. pylori-negative participants to explore potential differential correlations.

Among the 1,086 participants included in the final analysis, a total of 313 deaths occurred during the follow-up period (28.8%). RCS analyses were applied to examine potential nonlinear relationships between urinary metal concentrations and all-cause mortality, modeled as continuous variables (Figs. 5, 6 and 7).

Fig. 5.

Restricted cubic spline (RCS) curves showing the association between urinary Lead concentrations and all-cause mortality, stratified by H. pylori infection status. Panels depict adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality risk in (A) H. pylori-positive and (B) H. pylori-negative participants. Models were adjusted for demographic, clinical, and biochemical covariates including co-exposure to other metals. Four knots were placed at the 5th, 35th, 65th, and 95th percentiles

Fig. 6.

Restricted cubic spline (RCS) curves for the association between urinary Thallium concentrations and all-cause mortality, stratified by H. pylori infection status. Dose–response associations are presented for (A) H. pylori-positive and (B) H. pylori-negative individuals. Four knots were placed at the 5th, 35th, 65th, and 95th percentiles

Fig. 7.

Restricted cubic spline (RCS) analysis of urinary Tungsten concentrations and all-cause mortality risk, by H. pylori status. A H. pylori-positive and (B) H. pylori-negative participants. Four knots were placed at the 5th, 35th, 65th, and 95th percentiles

RCS analyses further explored the potential dose–response relationships between urinary metals and all-cause mortality across H. pylori subgroups. In Fig. 6, urinary Thallium demonstrated a significant association with all-cause mortality in both H. pylori-positive (P for nonlinearity = 0.2731) and H. pylori-negative individuals (P for nonlinearity = 0.0127), with evidence of nonlinearity in the latter group. In contrast, Fig. 7 shows that urinary Tungsten was significantly associated with increased mortality among H. pylori-positive participants (P overall = 0.0074), particularly at lower exposure levels, whereas no significant association was observed in the H. pylori-negative subgroup (P overall = 0.1268). No meaningful nonlinear association was identified for urinary lead in either subgroup (Fig. 5).

Stratified analyses were conducted separately in each subgroup, with 144 deaths in the H. pylori-positive group and 169 in the H. pylori-negative group. Kaplan–Meier survival curves revealed that higher urinary Thallium concentrations were associated with lower all-cause mortality in both H. pylori-positive (P < 0.0001) and H. pylori-negative (P = 0.0036) participants (Fig. 3). However, this inverse relationship did not remain significant after adjustment for covariates in Cox regression models (Table 5). Urinary lead was associated with increased mortality only in crude analyses among H. pylori-negative participants, and this association disappeared after multivariable adjustment (Table 4).

Notably, for urinary Tungsten, higher concentrations (Q4 vs. Q1) were significantly associated with increased all-cause mortality in H. pylori-positive participants, even after adjusting for demographic, clinical, and co-exposure variables (HR = 3.129, 95% CI: 1.031–9.493, Table 6). No significant association between urinary Tungsten and mortality was observed in H. pylori-positive individuals after adjustment when analyzed in quartiles. Notably, for urinary Tungsten, higher concentrations (Q4 vs. Q1) were significantly associated with increased all-cause mortality in H. pylori-negative participants. In the fully adjusted model (Model 2), the hazard ratio was 3.054 (95% CI: 1.083–8.608; P = 0.027; Table 6), indicating that individuals in the highest quartile of urinary Tungsten had more than three folds the risk of death compared to those in the lowest quartile. The lower bound of the confidence interval exceeding 1.0 supports the clinical and statistical significance of this finding. The details of FDR correction using BH method of urinary metals and all-cause mortality among H. pylori positive and negative were shown in supplementary Table 3 and 4.

These findings suggest a potential modifying effect of H. pylori infection status on the relationship between Tungsten exposure and mortality. Tungsten was the only metal among those evaluated that remained independently associated with mortality in H. pylori-positive individuals after full covariate adjustment.

Sensitivity analyses of E-values on the urinary metals and H. pylori infection and all-cause mortality

To evaluate the potential influence of unmeasured confounding, we conducted sensitivity analyses using E-values for the associations between urinary metals and both H. pylori infection and all-cause mortality. For H. pylori infection, the E-values were 1.534 for urinary lead, 5.948 for urinary thallium, and 5.456 for urinary tungsten (Supplementary Table 5). For all-cause mortality among H. pylori-positive participants, the E-values were 2.000 for lead, 11.599 for thallium, and 5.710 for tungsten (Supplementary Table 6). Among H. pylori-negative participants, the corresponding E-values were 1.505 for lead, 13.369 for thallium, and 2.864 for tungsten (Supplementary Table 7).

Discussion

Our study examined the association between urinary metal exposure and H. pylori infection, as well as the overall mortality in cohorts with and without H. pylori infection, using data spanning the NHANES 1999–2000. Our findings are consistent with and extend prior work by Krueger and Wade [11], who were among the first to report positive associations between heavy metal exposure and H. pylori seropositivity. Unlike their study, which focused solely on lead and cadmium, we evaluated a wider panel of metals and incorporated mortality follow-up, thereby offering a more comprehensive assessment of health risks. Our findings revealed a positive association between urinary Tungsten levels and H. pylori seropositivity. However, only in H. pylori-negative participants, urinary Tungsten was positively associated with all-cause death.

It is important to note that due to the cross-sectional design of the NHANES data, no temporal or causal inference can be made. The mechanistic hypotheses discussed below are speculative and require validation in prospective or experimental settings. Exposure to certain metals has been associated with cancer and various diseases, including gastritis and even gastric cancer [29–31]. Specifically, heavy metals can decrease mucosal thickness and mucus content, as well as influence basal acid output, affecting the function of E-cadherin and inducing damage through the generation of ROS [32, 33]. On the other hand, heavy metals, whether through direct or indirect mechanisms, have been implicated in the induction of ROS generation, contributing to gastric mucosal and DNA lesions. These molecular disruptions, in turn, have cascading associations on gene regulation, signal transduction pathways, and cell growth. The cumulative impact of these processes ultimately leads to carcinogenesis in the gastric tissue [34]. These findings indicate a positive association between heavy metal exposure and H. pylori infection. The relationships of heavy metal exposure on the gastric environment, including alterations in mucosal thickness, acid output, and the generation of ROS, may create conditions favorable for H. pylori infection.

The robust association observed between H. pylori seropositivity and exposures to Lead and Thallium in the US population may be indicative of an increased vulnerability or susceptibility, aligning with findings from prior studies [11, 35]. This heightened susceptibility could be attributed to factors such as increased and faster absorption of these metals, especially during periods when hygienic conditions may make individuals more likely to be exposed to H. pylori. The relationships between Thallium exposure and human health, along with the underlying mechanisms, remain unclear. An investigation discovered that Thallium treatment in HEK293 cells resulted in reduced protein synthesis, impaired ribosome biogenesis, and blocked cell cycle progression and apoptosis [36]. Moreover, Thallium exposure has been associated with early-onset mitochondrial dysfunction, alterations in transmembrane mitochondrial potential, changes in mitochondrial ROS levels, and fluctuations in ethanol and lactate contents [37, 38].

Interestingly, we observed a significant association between urinary tungsten and mortality among H. pylori-positive individuals only when tungsten was modeled as a continuous variable, whereas quartile-based analysis did not show consistent associations. In contrast, among H. pylori-negative individuals, higher tungsten quartiles were consistently associated with increased mortality risk, suggesting a potential linear dose–response relationship. In H. pylori-negative individuals, the linear association may reflect a more direct toxic effect of tungsten in the absence of underlying gastric inflammation. Tungsten can substitute for molybdenum in essential metalloenzymes, thereby disrupting redox homeostasis and mitochondrial function, promoting ROS production, and inducing systemic oxidative stress [39, 40]. These toxic effects may account for the observed increased mortality. Conversely, in H. pylori-positive individuals, pre-existing inflammation and oxidative stress from the infection [41, 42], may interact with or obscure the effects of tungsten, leading to a non-linear exposure–response pattern that is only detectable in continuous models. A potential synergistic effect is also plausible, where tungsten exacerbates H. pylori-related mucosal damage and inflammatory responses, collectively increasing mortality risk. Although prior studies suggest oral tungsten may reduce inflammation by inhibiting certain gut bacteria such as Enterobacteriaceae and Escherichia coli [43, 44], its role in modulating H. pylori survival remains unclear. Given tungsten's interference with molybdenum-dependent microbial and host enzymes [45, 46], further mechanistic research is needed to understand its dual effects on infection dynamics and host pathology [3].

Moreover, our study did not establish clear dose–response relationships for most metals, and the nonlinear associations observed in spline models may reflect residual confounding or threshold effects rather than biologically plausible gradients. Given the complexity of metal–host–microbiome interactions and the cross-sectional nature of our exposure assessment, caution is warranted in interpreting the biological plausibility of our findings. Thus, while these hypotheses provide a conceptual framework, our results should primarily be viewed as observational associations requiring further mechanistic validation. From a clinical and public health perspective, the observed associations between urinary Tungsten levels and both H. pylori seropositivity and all-cause mortality suggest that environmental metal exposure may serve as a novel, underrecognized contributor to gastrointestinal and systemic health risks. These findings underscore the importance of incorporating environmental toxicant assessment into broader risk stratification frameworks, particularly in populations with limited socioeconomic resources who may be disproportionately exposed. Although urinary metal levels are not yet part of routine clinical screening, our results raise the possibility that Tungsten could serve as a biomarker for identifying vulnerable individuals at elevated mortality risk, especially those without known H. pylori infection.

Moreover, these findings may carry implications for future prevention strategies. Given the widespread industrial and environmental presence of Tungsten, targeted public health interventions—such as environmental surveillance, regulation of industrial emissions, and community-level screening—may help mitigate exposure-related health consequences. In high-risk subgroups, efforts to monitor environmental exposures and address modifiable co-exposures could serve as a complement to traditional clinical approaches for reducing gastrointestinal disease burden and mortality.

To our knowledge, this is the first study to examine the relationship between urinary metal exposure and H. pylori infection, offering new insights that contribute to the existing literature. This research has the potential to refine risk stratification and guide personalized approaches to healthcare for individuals with H. pylori infection and further drug targets, ultimately contributing to improved patient outcomes and enhanced geriatric care.

Limitations

This study has several limitations. First, NHANES is cross‑sectional and cannot establish causality; although we linked mortality follow‑up to enhance clinical relevance, analyses were restricted to the 1999–2000 cycle—the only wave with concurrent H. pylori serology and urinary metals—limiting sample size and temporal/generalizability as exposure patterns and demographics have changed. Second, single spot urine samples may not accurately reflect chronic metal exposure due to short biological half-lives, diurnal variation, and inter-individual differences in excretion patterns. This exposure misclassification could lead to attenuation of true associations or false positive findings. Third, although we adjusted for a wide range of covariates, residual confounding is likely due to unmeasured factors such as dietary intake, occupational exposure, geographic variation, and additional renal function indicators. These factors may influence both metal levels and health outcomes, and their absence may bias the observed associations. In addition, due to reduced sample size in stratified models, the power to detect subgroup-specific associations may be limited. As such, the results of these subgroup analyses should be interpreted with caution and considered hypothesis-generating rather than conclusive. What’s more, we acknowledge that our sensitivity analyses were limited in scope. While we conducted E-value analyses to evaluate the robustness of associations to unmeasured confounding and applied FDR correction to account for multiple comparisons, we did not perform more advanced sensitivity analyses such as alternative exposure categorizations, use of continuous versus categorical modeling strategies, or varying covariate adjustment sets. The absence of such analyses represents a methodological limitation and may affect the certainty and generalizability of our findings. In addition, we did not perform formal power calculations, and our relatively small sample size, particularly in subgroup analyses stratified by H. pylori infection status, may have resulted in limited statistical power to detect modest associations. This limitation raises the possibility that some null findings may reflect insufficient power rather than true absence of effect. Future studies with larger sample sizes and adequately powered subgroup analyses are warranted to confirm these findings. Lastly, the proposed biological mechanisms are speculative, with limited human evidence, and the findings may not generalize beyond the U.S. population. Future longitudinal studies with more comprehensive exposure assessments and covariate data are needed to validate and clarify these associations.

Conclusion

Our study identified a significant association between urinary Tungsten concentrations and both H. pylori infection and all-cause mortality in a nationally representative sample of U.S. adults from the NHANES 1999–2000. Notably, urinary Tungsten was significantly associated with mortality in H. pylori-positive individuals when modeled as a continuous variable, but not in categorical analyses. Conversely, a dose–response relationship was observed only in H. pylori-negative individuals when tungsten was categorized into quartiles. Interventions aimed at reducing Tungsten exposure may represent a novel avenue for improving health outcomes in vulnerable populations.

Abbreviations

BMI

Body mass index

CI

Confidence interval

CKD-EPI

Chronic Kidney Disease Epidemiology Collaboration Equation

DBP

Diastolic blood pressure

ELISA

Enzyme-linked immunosorbent assay

eGFR

Estimated Glomerular Filtration Rate

HR

Hazards ratio

H. pylori

Helicobacter pylori

IgG

Immunoglobulin G

NHANES

National Health and Nutrition Examination Survey

OD

Optical density

OR

Odds ratio

RCS

Restricted cubic spline

ROS

Reactive oxygen species

SBP

Systolic blood pressure

Authors’ contributions

XDM, SY conceived and designed the study, acquired the data and drafted the manuscript; YJX analyzed the data, and contributed to the interpretation of the results and critical revision of the manuscript for important intellectual content; YJX and XDM developed the software and provided technical support. SY had the primary responsibility for final content. All authors have read and approved the final manuscript. The authors reported no conflicts of interest.

Funding

None.

Data availability

The raw data supporting the conclusions of this article can be found here: https://www.cdc.gov/nchs/nhanes/.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). Consent to participate was obtained and the National Center for Health Statistics (NCHS) ethics committee approved the NHANES study protocol. Study protocols for the NHANES were approved by the NCHS ethnics review board (Protocol #2011–17). All information from the NHANES program is available and free for the public, so the agreement of the medical ethics committee board was not necessary.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.