Associations Between Domestic Water Hardness, Genetic Risk and Severe MASLD: A Large‐Scale Cohort Study

Xiaoyu Zhao; Zihao Gui; Houde He; Mengyuan Yang; Yijun Lin; Jie Shen; Ningjian Wang; Heng Wan; Feng Tian

doi:10.1096/fj.202601814RR

. 2026 Aug 1;40(15):e72179. doi: 10.1096/fj.202601814RR

Associations Between Domestic Water Hardness, Genetic Risk and Severe MASLD: A Large‐Scale Cohort Study

Xiaoyu Zhao ¹, Zihao Gui ², Houde He ³, Mengyuan Yang ¹, Yijun Lin ¹, Jie Shen ^2,⁴, Ningjian Wang ⁵, Heng Wan ^2,^4,^✉, Feng Tian ^1,^✉

PMCID: PMC13428296 PMID: 42541335

ABSTRACT

This study aims to investigate the association between domestic water hardness and the risk of severe metabolic dysfunction‐associated steatotic liver disease (MASLD) and to assess the potential modifying role of genetic susceptibility. The UK Biobank is a prospective cohort study that recruited over 500,000 participants across the United Kingdom between 2006 and 2010. Domestic water hardness was assessed based on calcium carbonate (CaCO₃) concentration. Severe MASLD was defined using ICD‐10 codes and death registry data. A polygenic risk score (PRS) for severe MASLD was constructed using 61 single nucleotide polymorphisms (SNPs) and divided into tertiles (low, intermediate, high risk). Cox proportional hazards regression models and restricted cubic spline analyses were employed. A total of 7039 incident severe MASLD cases (1.5%) were documented during a mean 13.8‐year follow‐up. Compared to water hardness < 100 mg/L, individuals with domestic water hardness of 100–200 mg/L were associated with a 22% lower risk of severe MASLD (HR 0.78, 95% CI 0.72–0.86), while water hardness > 200 mg/L was associated with an 11% higher risk (HR 1.11, 95% CI 1.05–1.16). A non‐linear relationship was observed between water hardness and severe MASLD incidence (p for non‐linear < 0.001). The association did not significantly differ across strata of genetic susceptibility (p‐interaction > 0.05). The study revealed a U‐shaped relationship between water hardness and severe MASLD, which was robust to genetic risk. An optimal water hardness range (100–200 mg/L) may serve as a modifiable environmental factor for the prevention of severe MASLD.

Keywords: calcium carbonate, domestic water hardness, genetic risk, polygenic risk score, severe MASLD, UK Biobank

This graphical abstract summarizes the association between domestic water hardness and severe MASLD risk using UK Biobank data. Water hardness was assessed by calcium carbonate concentration via postal codes. Participants were categorized into low, moderate, and high genetic risk groups based on polygenic risk score tertiles. Cox proportional hazards models revealed a nonlinear relationship: compared with water hardness < 100 mg/L, moderate hardness (100–200 mg/L) reduced severe MASLD risk by 22% (HR 0.78, 95% CI 0.72–0.86), whereas high hardness (> 200 mg/L) increased risk by 11% (HR 1.11, 95% CI 1.05–1.16). No significant interaction with genetic predisposition was observed.

graphic file with name FSB2-40-e72179-g003.jpg

1. Introduction

Metabolic dysfunction‐associated steatotic liver disease (MASLD) refers to liver diseases where hepatocyte macrovesicular steatosis is ≥ 5% after excluding specific causes such as excessive alcohol consumption [1, 2], and has become one of the main causes of chronic liver diseases worldwide. The disease spectrum progresses gradually, covering from simple fatty liver, non‐alcoholic fatty liver disease, liver fibrosis to cirrhosis [1, 2]. Severe MASLD is specifically defined as hospitalization or mortality caused by MASLD/MASH, exhibiting more severe clinical manifestations and outcomes [3, 4]. Notably, while lifestyle interventions remain the cornerstone of severe MASLD management [5], the potential impact of domestic water, an important component of the daily diet, on severe MASLD has not been fully investigated.

The impact of the hardness of domestic water on health has received increasing attention. Continuous research on hard water has been carried out, and it has been found to be associated with a variety of diseases, such as eczema [6], atrial fibrillation [7], and other fields. Hard water refers to water that contains a high amount of soluble calcium and magnesium compounds. Although studies have confirmed that high concentrations of minerals in water can affect cardiovascular health, the direct impact of these minerals on severe MASLD still requires further exploration [8]. In recent years, some research has begun to focus on the association between different water sources and severe MASLD, such as analyses of deep‐sea seawater [9], hydrogen‐rich water [10], and high plain water intake [11]. However, the research on domestic water hardness and severe MASLD is lacking, which suggests the necessity of further relevant research.

Increasing evidence indicates that severe MASLD is a complex disease caused by the interaction of genetic susceptibility, environmental exposure, and lifestyle factors such as high‐calorie diet and lack of exercise [12, 13]. The Genome‐Wide Association Study (GWAS) conducted on individuals of European ancestry identified multiple risk loci significantly associated with severe MASLD, revealing the important genetic background of this disease [12]. However, it remains unclear how genetic variants interact with domestic water hardness to influence disease risk and severity.

This study systematically explored the association between household water hardness and the risk of severe MASLD using the UK Biobank, and further analyzed the regulatory role of genetic risk in this association. By revealing the interactive effects of environmental exposure and genetic background on disease occurrence, the research results not only provide a new perspective for understanding the pathogenic mechanism of severe MASLD, but also offer a scientific basis for formulating targeted public health intervention strategies and water quality management measures.

2. Materials and Methods

2.1. Study Design and Participants

The UK Biobank adopted a long‐term prospective follow‐up design, established connections with institutions such as the National Health Service of the UK, and continuously tracked the health records of volunteers. Over 500 000 volunteers were recruited from 2006 to 2010, average follow‐up period was 13.8 years and the follow‐up period ended on December 19, 2022 [14]. For more information about UK Biobank, please refer to previous publications [14] or visit its official website (http://www.ukbiobank.ac.uk). As shown in the research flowchart (Figure 1), among the initial 502 269 participants, 46 209 were excluded due to concurrent liver diseases, excessive alcohol consumption, and lack of data on water hardness. Among them, the definition of excessive alcohol consumption is based on the clinical diagnostic codes for alcohol use disorders recorded in the hospital inpatient records and primary care data associated with the UK Biobank (see Table S1). Subsequently, we further excluded 907 participants who had been diagnosed with severe MASLD within the previous two years of the baseline survey or at the start of the study. Ultimately, 455 153 participants were included in the analysis cohort.

Flowchart of the study participants. MASLD, metabolic dysfunction‐associated steatotic liver disease.

2.2. Domestic Water Hardness

Domestic water hardness data includes the concentrations of calcium carbonate (CaCO₃), calcium, and magnesium. Based on participants' residential addresses, historical water quality data were obtained from local water supply companies in England, Wales, and Scotland, with postal codes matched to water supply zones (location rounded to the nearest kilometer). Exposure assessment used 2005 data, with water samples collected quarterly or semi‐annually to provide annual mean values for each postal code area. Specific analytical methods were not detailed in public documents. Water hardness was primarily assessed as calcium carbonate concentration, representing the total dissolved calcium and magnesium ions, expressed as calcium carbonate equivalents [6, 7, 15]. This study used two classification systems based on CaCO₃ levels to characterize domestic water hardness: (i) as a continuous variable (in 50 mg/L increments); (ii) as a categorical variable following the World Health Organization (WHO) classification (soft water < 100 mg/L, moderately hard water ≥ 100 to ≤ 200 mg/L, hard water > 200 mg/L) [16]. Additionally, calcium and magnesium concentrations were quantified and divided into quartiles to comprehensively assess the potential impact of mineral content on health outcomes.

2.3. Ascertainment of Incident Severe MASLD

Severe MASLD was defined as a composite outcome of hospitalization or death due to fatty liver or steatohepatitis, based on the framework of Petermann‐Rocha et al. [17], and identified using ICD‐10 codes and death registry data. It corresponds to ICD‐10 codes K76.0 [fatty (liver change), not elsewhere classified] and K75.8 [non‐alcoholic steatohepatitis and other specified inflammatory liver diseases]. The death database includes data fields 40 001 [primary cause of death] and 40 002 [contributory/secondary cause of death]. Diagnosis dates were determined via linked hospital/death databases or the last available follow‐up date. This definition was adapted from previous studies.

2.4. Genetic Risk Score for Severe MASLD

The basic principle of GWAS is to compare the genotypes of patient groups and healthy control groups on a large scale and identify the genetic markers that are more common in patients. Based on the genetic loci identified by GWAS, we referred to previous literature and selected single nucleotide polymorphisms (SNPs) related to severe MASLD, and used them to construct the corresponding polygenic risk score (PRS). This PRS contains 61 SNPs (details see Table S2), and the minor allele frequencies of these SNPs in severe MASLD are all greater than 0.05 [12, 13, 18]. In our study, the PRS was further categorized into low, medium, and high genetic risk groups based on the tertiles for analysis.

2.5. Covariates

In this study, the definition of covariates is mainly based on the baseline measurement data from the UK Biobank and previous studies [17, 19]. Age is calculated based on the date of birth; gender, race, educational level, smoking status, and drinking status are derived from self‐report at the baseline. Body mass index (BMI) was calculated as weight/height² (kg/m²) and classified as underweight (< 18.5), normal (18.5–24.9), overweight (25–29.9), or obese (≥ 30). The Townsend Deprivation Index (TDI) reflects area‐level socioeconomic deprivation, calculated from residential postcodes, with higher scores indicating greater deprivation. Physical activity was quantified as weekly metabolic equivalent (MET) minutes from the baseline questionnaire, and participants were categorized into low (< 600 MET‐min/week), moderate (600–< 3000 MET‐min/week), and high (≥ 3000 MET‐min/week) activity levels. For components of metabolic syndrome, central obesity (waist circumference ≥ 94 cm in men/≥ 80 cm in women), hyperglycemia/diabetes (fasting glucose ≥ 5.6 mmol/L or diagnosed diabetes), hypertension (blood pressure ≥ 130/85 mmHg or diagnosed hypertension), low High‐Density Lipoprotein (HDL) cholesterol (< 1.03 mmol/L in men/< 1.29 mmol/L in women), and hypertriglyceridemia (≥ 1.7 mmol/L). A directed acyclic graph (DAG) was constructed using DAGitty (v3.1) to identify potential confounders and elucidate causal pathways (Figure S1) [20, 21]. Stepwise analysis was used to validate the rationale for covariate adjustment (Table S3). For covariates with < 10% missing data, we imputed missing values by method: continuous variables (mean for normal, median for non‐normal distributions) and categorical variables (missing indicator method).

2.6. Statistical Analysis

We grouped the participants based on whether they had severe MASLD, and presented the distribution of water hardness between the two groups using a bar chart. Furthermore, baseline demographic characteristics were summarized across WHO water hardness categories to assess their distributional differences by exposure level. One‐way ANOVA or Kruskal Wallis test for continuous variables; Pearson chi‐square test for categorical variables. We established a multivariate Cox proportional hazards model to assess the risk of severe MASLD under different levels of household water hardness: Model 1 adjusted for age, gender, race, BMI, TDI, educational level, physical exercise, smoking and drinking status; Model 2 further adjusted for components of metabolic syndrome. We used restricted cubic splines (RCS) to analyze the non‐linear association between water hardness and disease risk, with nodes set at the 10th, 50th, and 90th percentiles.

We divided PRS into three genetic risk categories (low, medium, and high) to analyze the impact of genetic susceptibility on the incidence of severe MASLD. We performed stratified analyses by sex, age, race, alcohol consumption, TDI, socioeconomic status (SES; assessed at the individual level via latent class analysis based on household income, education, and employment status, categorized as high, medium, or low) [22], dietary quality (measured by HEI‐2015 score) [23], physical activity, BMI, fatty liver index (FLI), fibrosis‐4 index (FIB‐4), geographic region, metabolic phenotype, and cardiovascular disease [5]. FIB‐4 was calculated using the standard formula [24]: FIB‐4 = (age × AST)/(platelets × √ALT), with AST (aspartate aminotransferase), ALT (alanine aminotransferase), and platelet counts from baseline blood samples and age at baseline. Interaction effects were assessed using the likelihood ratio test.

To ensure the reliability of the research conclusions, we conducted the following sensitivity analyses: (a) Re‐classifying water hardness based on the standards of the United States Geological Survey (USGS): soft water (≤ 60 mg/L), moderately hard water (> 60, ≤ 120 mg/L), hard water (> 120, ≤ 180 mg/L), and very hard water (> 180 mg/L) [7, 15]; (b) Exclude participants with less than 10 years of residence to re‐examine the relationship between water hardness and severe MASLD, thereby reducing the influence of residence changes on the results; (c) To evaluate potential effect modification by water intake, we further adjusted for water intake in the main model, stratified by intake categories (low 0.5–3, medium 4–6, high 7–10 cups/day), and tested the multiplicative interaction using the likelihood ratio test. Daily water intake was collected via the UK Biobank baseline questionnaire (“How many cups of water do you drink per day?”). Responses of “< 1 cup” were coded as 0.5 cups/day. In sensitivity analyses, missing values and extremes (0 or > 10 cups/day) were excluded. Final water intake was treated as a categorical variable (0.5 and 1–10 cups/day). (d) The impact of the excessive alcohol consumption definition was assessed by excluding participants who reported drinking daily/almost daily or 3–4 times per week at baseline and re‐examining the association. (e) To explore whether systemic calcium homeostasis mediates the U‐shaped association between water hardness and severe MASLD, we analyzed baseline serum calcium. Serum calcium was albumin‐corrected (corrected calcium (mmol/L) = measured calcium + 0.02 × (40 − albumin)), and multivariate linear regression was used to assess its association with WHO water hardness categories, adjusting for age, sex, race, BMI, estimated glomerular filtration rate (eGFR), serum 25‐hydroxyvitamin D, and socioeconomic and lifestyle factors. (f) To further explore potential biological pathways, we compared the serum levels of ALT, AST, and CRP among WHO water hardness groups. The inter‐group differences were analyzed using the Kruskal‐Wallis test and the post hoc Dunn test corrected by Bonferroni. Subsequently, a multivariate linear regression model was fitted using the log‐transformed ALT, AST, and CRP as outcome variables, with covariates adjusted (age, gender, race, BMI, TDI, educational level, physical exercise, smoking and drinking status).

All statistical analyses were completed in R software (version 4.4.1). The p‐value was calculated using two‐sided hypothesis testing. The criterion for determining statistical significance was set at a p value less than 0.05.

3. Results

3.1. General Characteristics of the Participants

Table 1 presents the baseline characteristics of the study population stratified by severe MASLD status. Figure S2 presents the distribution of water hardness between the two groups in the form of a bar chart. Overall, among the 455 153 participants, the mean age was 57 years old. There were 7039 patients with severe MASLD (1.5%), who were older, had a similar sex distribution to the without severe group (52.6% vs. 54.5% female), and were more likely to be non‐white, have lower education, less physical activity, and lower socioeconomic status and income. They also had higher smoking rates, greater BMI, and a higher prevalence of metabolic syndrome components. Baseline characteristics stratified by WHO water hardness categories are presented in Table S4. Participants in hard water areas (> 200 mg/L) had a higher TDI (mean −0.87), indicating greater deprivation. The proportions of those with a university degree also differed across categories (standardized mean differences of 0.135).

TABLE 1.

Baseline characteristics of UK Biobank participants.

Variable	Total	Without severe MASLD	Severe MASLD	p
Socio‐demographics
Total n (%)	455 153 (100%)	448 114 (98.5%)	7039 (1.5%)	< 0.001
Baseline age (years), mean (SD)	56.6 (8.08)	56.56 (8.09)	57 (7.82)	< 0.001
Gender (female), n (%)	248 136 (54.5%)	244 437 (54.5%)	3699 (52.6%)	< 0.001
Deprivation index, mean (SD)	−2.09 [−3.61, 0.59]	−2.11 [−3.62, 0.56]	−1.11 [−3.11, 2.23]	< 0.001
Ethnicity, n (%)				< 0.001
White	430 421 (94.6%)	423 839 (94.6%)	6582 (93.5%)
Other racial groups	24 732 (5.4%)	24 275 (5.4%)	457 (6.5%)
BMI group				< 0.001
Underweight and normal weight	149 355 (32.8%)	148 616 (33.2%)	739 (10.5%)
Overweight	194 901 (42.8%)	192 351 (42.9%)	2550 (36.2%)
Obesity	110 897 (24.4%)	107 147 (23.9%)	3750 (53.3%)
Lifestyle
Smoking status, n (%)				< 0.001
Never	249 691 (54.9%)	246 499 (55%)	3192 (45.3%)
Previous	157 266 (34.6%)	154 448 (34.5%)	2818 (40%)
Current	48 196 (10.6%)	47 167 (10.5%)	1029 (14.6%)
Alcohol frequency intake, n (%)				< 0.001
Current	417 514 (91.7%)	411 451 (91.8%)	6063 (86.1%)
Never	21 206 (4.7%)	20 714 (4.6%)	492 (7%)
Previous	16 433 (3.6%)	15 949 (3.6%)	484 (6.9%)
Education				< 0.001
College or University degree	145 225 (31.9%)	143 736 (32.1%)	1489 (21.2%)
Other levels	309 928 (68.1%)	304 378 (67.9%)	5550 (78.8%)
MET group				< 0.001
Low physical activity	64 561 (14.2%)	63 309 (14.1%)	1252 (17.8%)
Moderate physical activity	246 775 (54.2%)	243 051 (54.2%)	3724 (52.9%)
High physical activity	143 817 (31.6%)	141 754 (31.6%)	2063 (29.3%)
Components of the metabolic syndrome
Central obesity (yes), n (%)	304 490 (66.9%)	298 137 (66.5%)	6353 (90.3%)	< 0.001
Hyperglycemia/diabetes (yes), n (%)	59 086 (13.0%)	57 494 (12.8%)	1592 (22.6%)	< 0.001
High blood pressure/hypertension (yes), n (%)	235 153 (51.7%)	230 246 (51.4%)	4907 (69.7%)	< 0.001
Low HDL (yes), n (%)	83 343 (18.3%)	80 916 (18.1%)	2427 (34.5%)	< 0.001
High triglycerides (yes), n (%)	170 904 (37.5%)	167 007 (37.3%)	3897 (55.4%)	< 0.001
Water hardness
Domesic water hardness in WHO				< 0.001
< 100 mg/L	265 828 (58.4%)	261 605 (58.4%)	4223 (60%)
100–200 mg/L	42 428 (9.3%)	41 898 (9.3%)	530 (7.5%)
> 200 mg/L	146 897 (32.3%)	144 611 (32.3%)	2286 (32.5%)
Domestic water hardness in USGS				< 0.001
≤ 60 mg/L	178 936 (39.3%)	176 144 (39.3%)	2792 (39.7%)
> 60, ≤ 120 mg/L	97 685 (21.5%)	96 077 (21.4%)	1608 (22.8%)
> 120, ≤ 180 mg/L	29 189 (6.4%)	28 860 (6.4%)	329 (4.7%)
> 180 mg/L	149 343 (32.8%)	147 033 (32.8%)	2310 (32.8%)
Calcium concentration				< 0.001
Q1 (≤ 15 mg/L)	113 789 (25.0%)	112 152 (25%)	1637 (23.3%)
Q2 (> 15, ≤ 42 mg/L)	113 788 (25.0%)	112 086 (25%)	1702 (24.2%)
Q3 (> 42, ≤ 99 mg/L)	113 788 (25.0%)	111 906 (25%)	1882 (26.7%)
Q4 (> 99 mg/L)	113 788 (25.0%)	111 970 (25%)	1818 (25.8%)
Magnesium concentraion				< 0.001
Q1 (≤ 2 mg/L)	113 789 (25.0%)	112 295 (25.1%)	1494 (21.2%)
Q2 (> 2, ≤ 4 mg/L)	113 788 (25.0%)	112 006 (25.0%)	1782 (25.3%)
Q3 (> 4, ≤ 5 mg/L)	113 788 (25.0%)	111 994 (25.0%)	1794 (25.5%)
Q4 (> 5 mg/L)	113 788 (25.0%)	111 819 (25.0%)	1969 (28.0%)

Open in a new tab

Note: Continuous variables are presented as mean (SD) or median (interquartile range), and categorical variables as numbers (percentages). p values were derived from analysis of variance (ANOVA) or the Kruskal‐Wallis test for continuous variables, and Pearson chi‐square test for categorical variables, comparing differences between severe MASLD and without severe MASLD groups. Education level was categorized as “College or University degree” and “Other levels”; “Other levels” includes basic education (A levels, AS levels, O levels, GCSEs, CSEs), vocational qualifications (NVQ, HND, HNC), and other professional qualifications. Components of metabolic syndrome were defined as: central obesity (waist circumference ≥ 94 cm in men, ≥ 80 cm in women); hyperglycemia (fasting glucose ≥ 5.6 mmol/L or diagnosed diabetes); hypertension (blood pressure ≥ 130/85 mmHg or diagnosed hypertension); low HDL cholesterol (< 1.03 mmol/L in men, < 1.29 mmol/L in women); and hypertriglyceridemia (≥ 1.7 mmol/L).

Abbreviations: BMI, body mass index; HDL, high‐density lipoprotein; MET, metabolic equivalent of task; USGS, United States Geological Survey; WHO, World Health Organization.

3.2. Associations of Water Hardness With Incident Severe MASLD

The results of the two adjusted Cox proportional‐hazards regression models showed consistent trends. The analysis revealed that for every one‐unit increase in the concentration of CaCO₃ in domestic water (continuous variable), the risk of severe MASLD increased (HR 1.05, 95% CI 1.03–1.08). According to the WHO classification, compared to the reference group, individuals with a concentration of 100–200 mg/L had a 22% lower risk of disease (HR 0.78, 95% CI 0.72–0.86), while those with a concentration > 200 mg/L had an 11% higher risk (HR 1.11, 95% CI 1.05–1.16). Additionally, the groups with the highest quartile levels of water calcium and water magnesium had a 25% (HR 1.25, 95% CI 1.17–1.34) and 45% (HR 1.45, 95% CI 1.35–1.55) increased risk of severe MASLD compared to the groups with the lowest quartile levels (Table 2).

TABLE 2.

The associations between different classifications of water hardness and the contents of calcium and magnesium in water with severe MASLD.

	N	Model 1 ^a	p	p for trend	Model 2 ^b	p	p for trend
CaCO₃_std	455 153	1.05 (1.03–1.08)	< 0.001		1.05 (1.03–1.08)	< 0.001
Ca²⁺_std	455 153	1.10 (1.07–1.12)	< 0.001		1.10 (1.07–1.12)	< 0.001
Mg²⁺_std	455 153	1.09 (1.06–1.11)	< 0.001		1.09 (1.06–1.11)	< 0.001
Ca_quartile				< 0.001			< 0.001
Q1 (≤ 15 mg/L)	113 789	Ref.			Ref.
Q2 (> 15, ≤ 42 mg/L)	113 788	1.15 (1.07–1.23)	0.001		1.07 (1.00–1.14)	0.006
Q3 (> 42, ≤ 99 mg/L)	113 788	1.26 (1.18–1.35)	< 0.001		1.20 (1.12–1.28)	< 0.001
Q4 (> 99 mg/L)	113 788	1.21 (1.13–1.29)	< 0.001		1.25 (1.17–1.34)	< 0.001
Mg_quartile				< 0.001			< 0.001
Q1 (≤ 2 mg/L)	113 789	Ref.			Ref.
Q2 (> 2, ≤ 4 mg/L)	113 788	1.32 (1.23–1.42)	< 0.001		1.28 (1.19–1.37)	< 0.001
Q3 (> 4, ≤ 5 mg/L)	113 788	1.36 (1.26–1.46)	< 0.001		1.33 (1.24–1.43)	< 0.001
Q4 (> 5 mg/L)	113 788	1.51 (1.41–1.62)	< 0.001		1.45 (1.35–1.55)	< 0.001
WHO classification
< 100 mg/L	265 828	Ref.		< 0.001	Ref.		0.002
100–200 mg/L	42 428	0.82 (0.75–0.90)	< 0.001		0.78 (0.72–0.86)	< 0.001
> 200 mg/L	146 897	1.04 (0.99–1.10)	< 0.001		1.11 (1.05–1.16)	< 0.001
USGS classification				< 0.001			0.006
≤ 60 mg/L	178 936	Ref.			Ref.
> 60, ≤ 120 mg/L	97 685	1.11 (1.04–1.18)	0.004		1.06 (0.99–1.12)	0.016
> 120, ≤ 180 mg/L	29 189	0.76 (0.68–0.85)	< 0.001		0.72 (0.64–0.80)	< 0.001
> 180 mg/L	149 343	1.07 (1.01–1.14)	< 0.001		1.12 (1.06–1.18)	< 0.001

Open in a new tab

Abbreviations: BMI, body mass index; HDL, high‐density lipoprotein; USGS, United States Geological Survey; WHO, World Health Organization.

^{^a}

Model 1: Adjusted for age, sex, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise).

^{^b}

Model 2: Adjusted for Model 1 plus the components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides).

3.3. Non‐Linear Associations Between Water Hardness and Incident Severe MASLD

To explore the potential non‐linear association between domestic water hardness and the risk of severe MASLD, we constructed a RCS curve. There was a significant nonlinear relationship between them (p for non‐linear < 0.001). The hazard ratio demonstrated a trend of first decreasing and then increasing with the increase in CaCO₃ concentration, with an increasing trend when exceeding 248.92 mg/L (Figure S3).

3.4. Associations of Water Hardness With Incident Severe MASLD Across Genetic Risk Status

Subgroup analysis (Figure 2) revealed no significant interaction between PRS categories and CaCO₃ levels (p = 0.235). Although the risk of severe MASLD was consistently lower at CaCO₃ concentrations of 100–200 mg/L across all genetic risk strata, this reduction appeared attenuated in the high genetic risk group. A similar attenuation pattern was observed under the USGS classification, with no significant interaction (p > 0.05; Figure S4).

Multivariable‐adjusted HRs (95% CIs) for genetic risk (continuous and quartiles) and MASLD (WHO classification of CaCO₃). HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). BMI, body mass index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, metabolic dysfunction‐associated steatotic liver disease; WHO, World Health Organization.

3.5. Stratification Analysis on the Associations of Water Hardness With Severe MASLD

Stratified analyses (Figures S5–S8) revealed significant interactions between water hardness and FIB‐4 index (p = 0.003) and race (p = 0.007). The protective effect of moderate water hardness (CaCO₃ 100–200 mg/L) was more pronounced among individuals with FIB‐4 < 1.3, with a 24% reduction in severe MASLD risk (HR 0.76, 95% CI 0.67–0.85), compared with a 15% reduction in the FIB‐4 ≥ 1.3 subgroup (HR 0.85, 95% CI 0.74–0.98), suggesting greater benefit in those without established liver fibrosis. For race, the inverse association was more evident in White participants (94.6% of the sample), with a 19% risk reduction (HR 0.81, 95% CI 0.74–0.89) at CaCO₃ 100–200 mg/L; however, this result should be interpreted cautiously due to limited statistical power in non‐White subgroups. Additionally, the largest risk reduction was observed in the least deprived group (HR 0.75, 95% CI 0.63–0.89). No significant interactions were detected in other subgroups, and individuals with CaCO₃ 100–200 mg/L consistently showed the lowest risk across all strata.

3.6. Sensitivity Analyses

In the sensitivity analysis, we conducted stratified analysis using the USGS water classification. Regardless of the subgroups, individuals with a CaCO₃ level among 120–180 mg/L (compared to those with a CaCO₃ level lower than 60 mg/L) consistently showed a lower risk of severe MASLD (Figures S9 and S10). Similarly, there were significant interactions in terms of race and FIB‐4. Secondly, the results after considering the stability of residence also showed a significant association between domestic water hardness and severe MASLD (all p values < 0.05) (Table S5). Thirdly, after further adjusting for water intake, the risk of severe MASLD decreased by 20% in individuals with a CaCO₃ level between 100 and 200 mg/L (HR 0.80, 95% CI 0.73–0.87, p < 0.001) (Table S6), and no significant interaction was observed in the stratified analysis based on water intake (p‐values for interaction > 0.05) (Figure S11). Fourth, sensitivity analyses excluding participants who reported drinking daily/almost daily or 3–4 times per week at baseline confirmed the significant association between water hardness and severe MASLD (all p < 0.05; Table S7). Fifth, after albumin correction, serum calcium differences across water hardness groups were statistically significant but clinically negligible. Compared with soft water, moderately hard water (100–200 mg/L) showed a marginally lower level (β = −0.0021 mmol/L), while hard water (> 200 mg/L) was slightly higher (β = 0.0039 mmol/L), both p < 0.05. However, all group means remained within 2.27–2.28 mmol/L, with between‐group differences (< 0.01 mmol/L) well below normal physiological variation (Tables S8 and S9). Sixth, Table S10 shows ALT, AST, and CRP levels across WHO water hardness groups. Median ALT and CRP were lower in the > 200 mg/L group than in the < 100 mg/L and 100–200 mg/L groups (both p < 0.05), with no significant difference between the 100–200 mg/L and < 100 mg/L groups. After multivariate adjustment (Table S11), ALT and CRP remained significantly lower in the > 200 mg/L group compared with the < 100 mg/L group (ALT: β = −0.0158, p < 0.001; CRP: β = −0.0478, p < 0.001), while AST showed no significant difference.

4. Discussion

As far as we know, this study is the first to systematically assess the association between domestic hard water and the risk of severe MASLD, and it has observed a significant nonlinear relationship between them. According to WHO criteria, individuals with CaCO₃ concentrations 100–200 mg/L exhibited significantly reduced risks of severe MASLD. Importantly, this association remained statistically significant across all genetic risk categories, although the effect was slightly attenuated in the high‐risk group.

Although no direct evidence has previously linked domestic water hardness to severe MASLD, this study is the first to identify a non‐linear U‐shaped association between the two. This pattern aligns with findings for other chronic diseases, including non‐linear relationships between water hardness and overall and site‐specific cancers (e.g., gastric and lung cancer) [15], as well as a lower risk of atrial fibrillation (AF) associated with moderate hardness (CaCO₃ 120–180 mg/L) [7]. The consistency with prior cardiovascular research on AF supports a biologically plausible hypothesis: domestic water hardness may exert non‐linear health effects across multiple endpoints of the cardiometabolic axis. Extensive cardiovascular evidence indicates that minerals in hard water may confer protection by modulating oxidative stress and chronic inflammation [7, 8]. Notably, severe MASLD shares common risk factors and pathophysiological mechanisms with cardiovascular disease, including insulin resistance and inflammation [25, 26, 27, 28], providing a theoretical basis for the protective association observed between water hardness and severe MASLD. Although the association strength is moderate, it is comparable to the effect sizes of other environmental exposures related to MASLD reported in the literature, such as air pollutants (odds ratio/risk ratio approximately 1.04–1.19) [29, 30], and exposure to mixtures of metals, phthalates, and polycyclic aromatic hydrocarbons in urine (odds ratio approximately 1.07–1.18) [31].

To our knowledge, this is the first study to explore the combined effect of domestic water hardness and genetic susceptibility on severe MASLD risk. The lowest risk was observed at a hardness of 100–200 mg/L, independent of genetic risk. Notably, the protective association of moderate water hardness against severe MASLD was attenuated in individuals at high genetic risk, with a consistent trend observed under the USGS classification. This suggests that, even in the absence of a statistically significant interaction, a high genetic risk background may partially diminish the potential benefits of hard water. Given that high genetic risk typically confers a stronger intrinsic drive toward lipid metabolism dysregulation, and that interaction tests are often underpowered, a non‐significant p value cannot exclude a true effect. These findings imply that high‐risk populations may require more proactive intervention strategies beyond reliance on environmental exposures. Nonetheless, this observation remains hypothesis‐generating and requires further validation in larger cohorts or more genetically homogeneous populations. Furthermore, stratified analyses by SES and TDI consistently confirmed the lowest risk within the 100–200 mg/L range, suggesting that socioeconomic factors may modify the association between water hardness and severe MASLD. Sensitivity analyses accounting for residential stability and personal water intake also supported the protective effect of this hardness range. The association did not differ significantly across other subgroups, indicating that appropriate water hardness may serve as a broadly applicable preventive measure. However, these secondary findings require further validation.

Water hardness exhibits a non‐linear relationship with severe MASLD risk. In moderately hard water (100–200 mg/L), the protective effect of minerals may be dominant. Studies have shown that the synergistic effect of minerals such as calcium, magnesium, and potassium in drinking water may mediate this protective effect [9]. Recent longitudinal studies and mechanistic investigations have suggested that MASLD is not merely a consequence of metabolic disorders, but may actively contribute to the vicious cycle of metabolic syndrome progression. This may occur through several pathways, including the secretion of pro‐inflammatory cytokines, the induction of systemic insulin resistance, and the disruption of adiponectin signaling [32, 33, 34]. These findings indicate that the relationship between metabolic syndrome components and MASLD progression may involve complex bidirectional interactions, and the precise causal direction remains to be fully elucidated. Given this, it is difficult to explain this association through metabolic pathways in this study. Moreover, the sensitivity analysis to investigate whether water hardness has a direct hepatotoxic effect showed that the ALT and CRP levels in the very hard water group were lower than those in the soft water group. This phenomenon may be related to the shift of the disease spectrum to more severe clinical phenotypes caused by the outcome of our study (hospitalization or death due to MASLD). However, the impact of extremely hard water on the risk of severe MASLD may involve the disruption of intracellular calcium homeostasis. Although the circulating calcium is strictly regulated and the differences in serum‐corrected calcium among different water hardness groups in this study are clinically negligible, the fluctuation range of intracellular calcium concentration is greater [35]. Theoretically, long‐term exposure to high calcium water may affect the intracellular calcium handling of liver cells through mechanisms that cannot be reflected by serum levels; however, this hypothesis has not been verified and requires further experimental research to explore. These direct pro‐fibrotic effects may dominate the overall effect on severe MASLD. Nevertheless, other explanations should also be considered. Extremely hard water may also reflect other geographically related water components (such as sodium, sulfate, and nitrate) or unmeasured regional environmental factors of common exposure. The WHO notes that soft water can corrode pipes, releasing harmful metals like lead into drinking water [16, 36]. Studies show that blood levels of lead and cadmium are linked to MASLD and liver fibrosis [37, 38]. Toxicological research also confirms that heavy metals can promote fat buildup, inflammation, and scarring in the liver [39]. Thus, soft water may contribute to MASLD through heavy metal toxicity or metabolic disruption. In conclusion, multiple pathways may lead to the non‐linear correlation between water hardness and severe MASLD. Moderate hardness may represent a favorable window for mineral exposure. These mechanisms require further experimental research and epidemiological investigations to be clarified.

This study found that higher water magnesium content was associated with an increased risk of severe MASLD. This is consistent with previous studies showing a positive correlation between water magnesium and the risk of MASLD [40] and cardiovascular diseases [7, 41, 42]. The association with water magnesium might reflect simultaneous exposure to other water components (such as sodium, sulfate) or unmeasured system‐related contaminants. However, multiple studies have shown that dietary magnesium intake is negatively correlated with the risk of MASLD [43, 44]. Its protective effect may stem from mechanisms such as improving insulin sensitivity in the liver and adipose tissue, anti‐inflammatory and antioxidant effects. A meta‐analysis concluded that continuous use of magnesium supplements for at least 4 months can significantly improve markers of insulin resistance regardless of whether diabetes is present [45]. In vitro experiments have shown that magnesium deficiency can exacerbate cellular oxidative damage [46]. The contradictory results may be due to differences in magnesium intake sources and assessment methods. Dietary magnesium is evaluated for long‐term intake through food frequency questionnaires (FFQ) [44, 47], while water magnesium is a population exposure indicator based on geographical regions. Moreover, compared to diet, drinking water typically contributes less to the daily magnesium requirement [16]. Inconsistencies in research findings may also be attributed to population‐specific factors.

This study has several limitations despite its large sample size, long follow‐up, and standardized data collection. First, water hardness exposure was assessed solely based on the baseline residential address, without accounting for prior residential mobility, workplace water hardness, or the use of bottled or filtered water. Although sensitivity analyses restricted to participants with ≥ 10 years of residence yielded consistent results, potential non‐differential exposure misclassification may have biased estimates toward the null. Second, water hardness data were collected before baseline, and exposure may have changed over time due to variations in water sources and treatment methods; additionally, the UK Biobank lacks data on other water contaminants (e.g., sodium, nitrate, metals, disinfection by‐products) and treatment methods. Third, severe MASLD was ascertained through hospitalization or death records, and the reference group may have included undiagnosed mild cases, likely resulting in conservative risk estimates. Fourth, exclusion of excessive alcohol consumption relied on clinical diagnostic codes rather than quantitative thresholds; although sensitivity analyses confirmed robustness, residual confounding cannot be fully excluded. Fifth, the UK Biobank does not currently provide serum or urinary magnesium data for the full cohort, precluding direct assessment of circulating mineral metabolism. Finally, the cohort is predominantly of White British ancestry, limiting generalizability to other ethnic groups, and the observational design cannot completely rule out residual confounding from unmeasured community‐level factors, precluding definitive causal inference.

This study offers valuable implications for public health practice. With the rising global prevalence of severe MASLD, adjusting domestic drinking water hardness presents a potential primary prevention strategy. Previous studies have shown that water hardness exceeding 180 mg/L is positively associated with certain cardiovascular and cerebrovascular diseases [7], all‐cause cancer [15], and kidney stones [48]. Therefore, the range of 100–200 mg/L may represent the optimal balance achieved between reducing the risk of severe MASLD and alleviating other related potential adverse effects. Given the various health impacts of water hardness, public health adjustments should be based on a comprehensive assessment of the overall benefits and risks. Therefore, we recommend piloting moderate mineralization of drinking water in areas with soft water and establishing a systematic monitoring mechanism. In areas with hard water, the water quality should be linked to the health outcomes of the population. This will provide a basis for formulating evidence‐based and economically efficient water resource management strategies in the future.

In conclusion, this prospective cohort analysis utilizing UK Biobank data demonstrates that exposure to optimal domestic water hardness levels (100–200 mg/L CaCO₃) is associated with the lowest risk of severe MASLD incidence. The observed U‐shaped nonlinear relationship suggests that domestic water hardness may serve as a modifiable environmental factor for disease prevention, with this regulatory effect remaining significant across different genetic risk strata. Given the high global prevalence of severe MASLD, these findings support further exploration of water hardness optimization as a population‐level strategy for reducing disease risk.

Author Contributions

X.Z. wrote the initial draft, made revisions and edits. Z.G. constructed the research methods, made revisions and edits. H.H., M.Y., Y.L. and J.S. made revisions and edits. N.W. made revisions and edits, supervised the project, conducted research. H.W. and F.T. made revisions and edits, supervised the project, conducted research, and obtained funding. All the authors have approved the final version of the paper and are responsible for its content.

Funding

This work was supported by Guangdong‐Hong Kong Technology Cooperation Funding Scheme (2024A0505040004), Guangdong Basic and Applied Basic Research Foundation (2023A1515140062), China Medical Foundation (2024CMFA06), Guangdong Provincial Higher Education Institutions' Characteristic and Innovative Projects (2022WTSCX009).

Ethics Statement

The UK Biobank was approved by the North West Multi‐Centre Research Ethics Committee (reference No. 16/NW/0274) in accordance with the Declaration of Helsinki, and all participants provided their informed consent.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: The directed acyclic graph based on causal hypothesis.

Figure S2: Water hardness classifications, calcium and magnesium quartiles across without severe MASLD and severe MASLD.

Figure S3: Restricted cubic spline for testing the hypothesis of non‐linear correlation between CaCO₃ and incident severe MASLD.

Figure S4: Multivariable‐adjusted HRs (95% CIs) for genetic risk (continuous and quartiles) and MASLD (USGS classification of CaCO₃). HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). BMI, body mass index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease.

Figure S5: Stratification analysis on the associations of water hardness (WHO classification) with incident severe MASLD. HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). In the analysis stratified by marked categorical variable, the corresponding variable was excluded from the model. FIB4, Fibrosis‐4 Index; FLI, Fatty Liver Index; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease.

Figure S6: Stratification analysis on the associations of water hardness (WHO classification) with incident severe MASLD. HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). In the analysis stratified by marked categorical variable, the corresponding variable was excluded from the model. HEI‐2015, Healthy Eating Index 2015; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease; MET, Metabolic Equivalent of Task.

Figure S7: Multivariable‐adjusted HRs (95% CIs) for SES (continuous and quartiles) and severe MASLD (WHO classification of CaCO₃). We employed latent class analysis, integrating three socio‐economic factors (family income, educational level, and employment status) to construct a comprehensive variable, Social Economic Status (SES), which was divided into three categories: low, medium, and high. HR of severe MASLD was adjusted for age, gender, race, BMI, lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). BMI, body mass index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease. SES, Social economic status.

Figure S8: Multivariable‐adjusted HRs (95% CIs) for the hypertension grade (continuous and quartiles) and severe MASLD (WHO classification of CaCO₃). According to the classification of hypertension levels, it was categorized as controlled, grade 1, grade 2, and grade 3 (with systolic blood pressure divided at 140, 160, and 180 mmHg, and diastolic blood pressure divided at 90, 100, and 110 mmHg). HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, low HDL, high triglycerides). BMI, body mass index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease.

Figure S9: Stratification analysis on the associations of water hardness (USGS classification) with incident severe MASLD. HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). In the analysis stratified by marked categorical variable, the corresponding variable was excluded from the model. BMI, body mass index; FIB4, Fibrosis‐4 Index; FLI, Fatty Liver Index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease.

Figure S10: Stratification analysis on the associations of water hardness (USGS classification) with incident severe MASLD. HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). In the analysis stratified by marked categorical variable, the corresponding variable was excluded from the model. HEI‐2015, Healthy Eating Index 2015; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease; MET, Metabolic Equivalent of Task.

Figure S11: Multivariable‐adjusted HRs (95% CIs) for the water intake (continuous and quartiles) and severe MASLD (WHO and USGS classification of CaCO₃). The amount of water intake was measured in cups. In analysis, we excluded participants whose self‐reported water intake at baseline was 0 cups, more than 11 cups, or had missing values. Group the water intake according to quartiles. HR of severe MASLD was adjusted for age, gender, race, BMI, socioeconomic status (Townsend deprivation index, education level), lifestyle behaviors (smoking status, alcohol intake, physical exercise), and components of the metabolic syndrome (central obesity, high glycaemia/diabetes, high blood pressure/hypertension, low HDL, high triglycerides). BMI, body mass index; HDL, High‐Density Lipoprotein; HR, hazard ratio; MASLD, Metabolic dysfunction‐associated steatotic liver disease.

Table S1: Disease definitions used in the UK Biobank study.

Table S2: Single‐nucleotide polymorphisms used to build the genetic risk score for MASLD.

Table S3: Association between water calcium carbonate concentration and severe MASLD: stepwise adjustment models.

Table S4: Baseline characteristics of UK Biobank participants by CaCO₃ category (using WHO).

Table S5: The association between different water hardness classifications and calcium and magnesium contents in water, and MASLD in participants with a residence duration of more than 10 years.

Table S6: After correcting water intake, the associations between different water hardness classifications and the calcium and magnesium contents in the water, as well as severe MASLD, were analyzed.

Table S7: Sensitivity analysis excluding participants with frequent alcohol consumption at baseline: association between water hardness, calcium and magnesium contents, and MASLD.

Table S8: Descriptive statistics of uncorrected and albumin‐corrected serum calcium by WHO water hardness categories.

Table S9: Multivariable linear regression associations of WHO water hardness categories with uncorrected and albumin‐corrected serum calcium.

Table S10: Serum levels of liver enzymes and CRP across WHO water hardness categories.

Table S11: Multivariable linear regression analysis of the association between WHO water hardness categories and log‐transformed serum ALT, AST, and CRP levels.

FSB2-40-e72179-s001.docx^{(14.7MB, docx)}

Acknowledgments

We are grateful to the participants of the UK Biobank study, the members of the survey teams, and the project development and management teams. This work was supported by the Guangdong‐Hong Kong Technology Cooperation Funding Scheme (2024A0505040004), the Guangdong Basic and Applied Basic Research Foundation (2023A1515140062), the China Medical Foundation (2024CMFA06), and the Guangdong Provincial Higher Education Institutions' Characteristic and Innovative Projects (2022WTSCX009). The funding body did not contribute to the study design, data collection, analysis, interpretation, report writing, or decision to submit the paper for publication. During the preparation of this work, the authors used DeepSeek V3 (DeepSeek Inc., accessed from January to June 2026) and BioIntelOS (DataSpeak, accessed from May to June 2026) for language editing and formatting assistance. After using these tools, the authors reviewed and edited the content as appropriate and take full responsibility for the content of the published article.

Contributor Information

Heng Wan, Email: wanhdr@163.com.

Feng Tian, Email: qqtina@smu.edu.cn.

Data Availability Statement

The data used in this study are from UK Biobank (Application Number 77740). Due to the data use agreement and privacy restrictions, the raw data are not publicly available. However, access can be requested through the UK Biobank Access Management System at https://www.ukbiobank.ac.uk/en/register‐apply. Code/analysis scripts are available upon reasonable request.

References

1. Maurice J. and Manousou P., “Non‐Alcoholic Fatty Liver Disease,” Clinical Medicine (London) 18, no. 3 (2018): 245–250, 10.7861/clinmedicine.18-3-245. [DOI] [PMC free article] [PubMed] [Google Scholar]
2. Powell E. E., Wong V. W., and Rinella M., “Non‐Alcoholic Fatty Liver Disease,” Lancet 397, no. 10290 (2021): 2212–2224, 10.1016/s0140-6736(20)32511-3. [DOI] [PubMed] [Google Scholar]
3. Chen L., Fan Z., and Lv G., “Associations of Muscle Mass and Grip Strength With Severe NAFLD: A Prospective Study of 333,295 UK Biobank Participants,” Journal of Hepatology 77, no. 5 (2022): 1453–1454, 10.1016/j.jhep.2022.05.005. [DOI] [PubMed] [Google Scholar]
4. Hagström H., Shang Y., Hegmar H., and Nasr P., “Natural History and Progression of Metabolic Dysfunction‐Associated Steatotic Liver Disease,” Lancet Gastroenterology & Hepatology 9, no. 10 (2024): 944–956, 10.1016/s2468-1253(24)00193-6. [DOI] [PubMed] [Google Scholar]
5. He P., Zhang Y., Ye Z., et al., “A Healthy Lifestyle, Life's Essential 8 Scores and New‐Onset Severe NAFLD: A Prospective Analysis in UK Biobank,” Metabolism 146 (2023): 155643, 10.1016/j.metabol.2023.155643. [DOI] [PubMed] [Google Scholar]
6. Lopez D. J., Singh A., Waidyatillake N. T., et al., “The Association Between Domestic Hard Water and Eczema in Adults From the UK Biobank Cohort Study,” British Journal of Dermatology 187, no. 5 (2022): 704–712, 10.1111/bjd.21771. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Tian F., Yu G., Yang M., et al., “Domestic Water Hardness, Genetic Risk, and Distinct Phenotypes of Cardiovascular Disease,” Environmental Health 24, no. 1 (2025): 9, 10.1186/s12940-025-01166-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Bykowska‐Derda A., Spychala M., Czlapka‐Matyasik M., Sojka M., Bykowski J., and Ptak M., “The Relationship Between Mortality From Cardiovascular Diseases and Total Drinking Water Hardness: Systematic Review With Meta‐Analysis,” Food 12, no. 17 (2023): 3255, 10.3390/foods12173255. [DOI] [PMC free article] [PubMed] [Google Scholar]
9. Lee C. Y. and Lee C. L., “Comparison of the Improvement Effect of Deep Ocean Water With Different Mineral Composition on the High Fat Diet‐Induced Blood Lipid and Nonalcoholic Fatty Liver Disease in a Mouse Model,” Nutrients 13, no. 5 (2021): 1732, 10.3390/nu13051732. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Kura B., Szantova M., LeBaron T. W., et al., “Biological Effects of Hydrogen Water on Subjects With NAFLD: A Randomized, Placebo‐Controlled Trial,” Antioxidants 11, no. 10 (2022), 10.3390/antiox11101935. [DOI] [PMC free article] [PubMed] [Google Scholar]
11. Wang X., Lin S., Gan S., et al., “Higher Plain Water Intake Is Related to Lower Newly Diagnosed Nonalcoholic Fatty Liver Disease Risk: A Population‐Based Study,” European Journal of Clinical Nutrition 75, no. 12 (2021): 1801–1808, 10.1038/s41430-021-00891-9. [DOI] [PubMed] [Google Scholar]
12. Anstee Q. M., Darlay R., Cockell S., et al., “Genome‐Wide Association Study of Non‐Alcoholic Fatty Liver and Steatohepatitis in a Histologically Characterised Cohort,” Journal of Hepatology 73, no. 3 (2020): 505–515, 10.1016/j.jhep.2020.04.003. [DOI] [PubMed] [Google Scholar]
13. Tang L., Li D., Ma Y., Cui F., Wang J., and Tian Y., “The Association Between Telomere Length and Non‐Alcoholic Fatty Liver Disease: A Prospective Study,” BMC Medicine 21, no. 1 (2023): 427, 10.1186/s12916-023-03136-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Allen N. E., Lacey B., Lawlor D. A., et al., “Prospective Study Design and Data Analysis in UK Biobank,” Science Translational Medicine 16, no. 729 (2024): eadf4428, 10.1126/scitranslmed.adf4428. [DOI] [PMC free article] [PubMed] [Google Scholar]
15. Yang H., Wang Q., Zhang S., Zhang J., Zhang Y., and Feng J., “Association of Domestic Water Hardness With All‐Cause and Cause‐Specific Cancers: Evidence From 447,996 UK Biobank Participants,” Environmental Health Perspectives 132, no. 6 (2024): 67008, 10.1289/ehp13606. [DOI] [PMC free article] [PubMed] [Google Scholar]
16. WHO Guidelines Approved by the Guidelines Review Committee , Guidelines for Drinking‐Water Quality: Fourth Edition Incorporating the First and Second Addenda (World Health Organization, 2022). [PubMed] [Google Scholar]
17. Petermann‐Rocha F., Gray S. R., Forrest E., et al., “Associations of Muscle Mass and Grip Strength With Severe NAFLD: A Prospective Study of 333,295 UK Biobank Participants,” Journal of Hepatology 76, no. 5 (2022): 1021–1029, 10.1016/j.jhep.2022.01.010. [DOI] [PubMed] [Google Scholar]
18. Jun D. W., “An Analysis of Polygenic Risk Scores for Non‐Alcoholic Fatty Liver Disease,” Clinical and Molecular Hepatology 27, no. 3 (2021): 446–447, 10.3350/cmh.2021.0133. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Petermann‐Rocha F., Wirth M. D., Boonpor J., et al., “Associations Between an Inflammatory Diet Index and Severe Non‐Alcoholic Fatty Liver Disease: A Prospective Study of 171,544 UK Biobank Participants,” BMC Medicine 21, no. 1 (2023): 123, 10.1186/s12916-023-02793-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Feeney T., Hartwig F. P., and Davies N. M., “How to Use Directed Acyclic Graphs: Guide for Clinical Researchers,” BMJ 388 (2025): e078226, 10.1136/bmj-2023-078226. [DOI] [PubMed] [Google Scholar]
21. Textor J., van der Zander B., Gilthorpe M. S., Liskiewicz M., and Ellison G. T., “Robust Causal Inference Using Directed Acyclic Graphs: The R Package ‘dagitty’,” International Journal of Epidemiology 45, no. 6 (2016): 1887–1894, 10.1093/ije/dyw341. [DOI] [PubMed] [Google Scholar]
22. Ye X., Wang Y., Zou Y., et al., “Associations of Socioeconomic Status With Infectious Diseases Mediated by Lifestyle, Environmental Pollution and Chronic Comorbidities: A Comprehensive Evaluation Based on UK Biobank,” Infectious Diseases of Poverty 12, no. 1 (2023): 5, 10.1186/s40249-023-01056-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
23. Lu Y. P., Xia B., Wang X. H., et al., “Healthy Dietary Patterns and the Incidence of Chronic Kidney Disease: Results From a Prospective Cohort Study,” BMC Public Health 25, no. 1 (2025): 511, 10.1186/s12889-025-21652-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
24. Sterling R. K., Lissen E., Clumeck N., et al., “Development of a Simple Noninvasive Index to Predict Significant Fibrosis in Patients With HIV/HCV Coinfection,” Hepatology 43, no. 6 (2006): 1317–1325, 10.1002/hep.21178. [DOI] [PubMed] [Google Scholar]
25. Benn M. and Nordestgaard B. G., “From Genome‐Wide Association Studies to Mendelian Randomization: Novel Opportunities for Understanding Cardiovascular Disease Causality, Pathogenesis, Prevention, and Treatment,” Cardiovascular Research 114, no. 9 (2018): 1192–1208, 10.1093/cvr/cvy045. [DOI] [PubMed] [Google Scholar]
26. Friedman S. L., Neuschwander‐Tetri B. A., Rinella M., and Sanyal A. J., “Mechanisms of NAFLD Development and Therapeutic Strategies,” Nature Medicine 24, no. 7 (2018): 908–922, 10.1038/s41591-018-0104-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
27. Ma W., Wu W., Wen W., et al., “Association of NAFLD With Cardiovascular Disease and All‐Cause Mortality: A Large‐Scale Prospective Cohort Study Based on UK Biobank,” Therapeutic Advances in Chronic Disease 13 (2022): 20406223221122478, 10.1177/20406223221122478. [DOI] [PMC free article] [PubMed] [Google Scholar]
28. Mantovani A., Csermely A., Petracca G., et al., “Non‐Alcoholic Fatty Liver Disease and Risk of Fatal and Non‐Fatal Cardiovascular Events: An Updated Systematic Review and Meta‐Analysis,” Lancet Gastroenterology & Hepatology 6, no. 11 (2021): 903–913, 10.1016/s2468-1253(21)00308-3. [DOI] [PubMed] [Google Scholar]
29. Cho S. H., Hwang S. E., Kim H. J., et al., “Association Between Long‐Term Exposure to Ambient Air Pollution and an Increased Risk of Steatotic Liver Disease,” Scientific Reports 15, no. 1 (2025): 21922, 10.1038/s41598-025-07495-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
30. Li F. R., Liao J., Zhu B., et al., “Long‐Term Exposure to Air Pollution and Incident Non‐Alcoholic Fatty Liver Disease and Cirrhosis: A Cohort Study,” Liver International 43, no. 2 (2023): 299–307, 10.1111/liv.15416. [DOI] [PubMed] [Google Scholar]
31. Li W., Xiao H., Wu H., et al., “Analysis of Environmental Chemical Mixtures and Nonalcoholic Fatty Liver Disease: NHANES 1999–2014,” Environmental Pollution 311 (2022): 119915, 10.1016/j.envpol.2022.119915. [DOI] [PubMed] [Google Scholar]
32. Byrne C. D., Armandi A., Pellegrinelli V., Vidal‐Puig A., and Bugianesi E., “Μetabolic Dysfunction‐Associated Steatotic Liver Disease: A Condition of Heterogeneous Metabolic Risk Factors, Mechanisms and Comorbidities Requiring Holistic Treatment,” Nature Reviews. Gastroenterology & Hepatology 22, no. 5 (2025): 314–328, 10.1038/s41575-025-01045-z. [DOI] [PubMed] [Google Scholar]
33. Hakeem A., van Berlo J., and Revelo X. S., “The Cardiohepatic Axis in Metabolic Disease: Liver to Heart,” JACC: Basic to Translational Science 10, no. 7 (2025): 101309, 10.1016/j.jacbts.2025.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
34. Zhou X. D., Lazarus J. V., Krittanawong C., et al., “Pharmacotherapy for Metabolic Dysfunction‐Associated Steatohepatitis: Heart‐Liver Co‐Management,” Lancet Gastroenterology & Hepatology 11, no. 6 (2026): 521–537, 10.1016/s2468-1253(25)00323-1. [DOI] [PubMed] [Google Scholar]
35. Papadopoulou A., Bountouvi E., and Karachaliou F. E., “The Molecular Basis of Calcium and Phosphorus Inherited Metabolic Disorders,” Genes (Basel) 12, no. 5 (2021): 734, 10.3390/genes12050734. [DOI] [PMC free article] [PubMed] [Google Scholar]
36. Leurs L. J., Schouten L. J., Mons M. N., Goldbohm R. A., and van den Brandt P. A., “Relationship Between Tap Water Hardness, Magnesium, and Calcium Concentration and Mortality due to Ischemic Heart Disease or Stroke in The Netherlands,” Environmental Health Perspectives 118, no. 3 (2010): 414–420, 10.1289/ehp.0900782. [DOI] [PMC free article] [PubMed] [Google Scholar]
37. Li Y., Liu Z., Chang Y., et al., “Associations of Multiple Toxic Metal Exposures With Metabolic Dysfunction‐Associated Fatty Liver Disease: NHANES 2011‐2018,” Frontiers in Nutrition 10 (2023): 1301319, 10.3389/fnut.2023.1301319. [DOI] [PMC free article] [PubMed] [Google Scholar]
38. Ma N., Bansal M. B., Chu J., Woodward M., and Branch A. D., “Heavy Metals Are Liver Fibrosis Risk Factors in People Without Traditional Liver Disease Etiologies,” Journal of Environmental Sciences 155 (2025): 329–342, 10.1016/j.jes.2024.08.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
39. Tinkov A. A., Aschner M., Santamaria A., et al., “Dissecting the Role of Cadmium, Lead, Arsenic, and Mercury in Non‐Alcoholic Fatty Liver Disease and Non‐Alcoholic Steatohepatitis,” Environmental Research 238 (2023): 117134, 10.1016/j.envres.2023.117134. [DOI] [PubMed] [Google Scholar]
40. He Q., Sun M., Huang J., et al., “Water Hardness and Digestive Diseases: A Large‐Scale Population‐Based Prospective Cohort Study,” International Journal of Surgery 112, no. 1 (2026): 922–934, 10.1097/js9.0000000000003506. [DOI] [PMC free article] [PubMed] [Google Scholar]
41. Morris R. W., Walker M., Lennon L. T., Shaper A. G., and Whincup P. H., “Hard Drinking Water Does Not Protect Against Cardiovascular Disease: New Evidence From the British Regional Heart Study,” European Journal of Cardiovascular Prevention and Rehabilitation 15, no. 2 (2008): 185–189, 10.1097/HJR.0b013e3282f15fce. [DOI] [PubMed] [Google Scholar]
42. Wodschow K., Villanueva C. M., Larsen M. L., et al., “Association Between Magnesium in Drinking Water and Atrial Fibrillation Incidence: A Nationwide Population‐Based Cohort Study, 2002–2015,” Environmental Health 20, no. 1 (2021): 126, 10.1186/s12940-021-00813-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
43. Da Silva H. E., Arendt B. M., Noureldin S. A., Therapondos G., Guindi M., and Allard J. P., “A Cross‐Sectional Study Assessing Dietary Intake and Physical Activity in Canadian Patients With Nonalcoholic Fatty Liver Disease vs Healthy Controls,” Journal of the Academy of Nutrition and Dietetics 114, no. 8 (2014): 1181–1194, 10.1016/j.jand.2014.01.009. [DOI] [PubMed] [Google Scholar]
44. Lu L., Chen C., Li Y., et al., “Magnesium Intake Is Inversely Associated With Risk of Non‐Alcoholic Fatty Liver Disease Among American Adults,” European Journal of Nutrition 61, no. 3 (2022): 1245–1254, 10.1007/s00394-021-02732-8. [DOI] [PubMed] [Google Scholar]
45. Simental‐Mendía L. E., Sahebkar A., Rodríguez‐Morán M., and Guerrero‐Romero F., “A Systematic Review and Meta‐Analysis of Randomized Controlled Trials on the Effects of Magnesium Supplementation on Insulin Sensitivity and Glucose Control,” Pharmacological Research 111 (2016): 272–282, 10.1016/j.phrs.2016.06.019. [DOI] [PubMed] [Google Scholar]
46. Afanas'ev I. B., Suslova T. B., Cheremisina Z. P., Abramova N. E., and Korkina L. G., “Study of Antioxidant Properties of Metal Aspartates,” Analyst 120, no. 3 (1995): 859–862, 10.1039/an9952000859. [DOI] [PubMed] [Google Scholar]
47. Emamat H., Ghalandari H., Totmaj A. S., Tangestani H., and Hekmatdoost A., “Calcium to Magnesium Intake Ratio and Non‐Alcoholic Fatty Liver Disease Development: A Case‐Control Study,” BMC Endocrine Disorders 21, no. 1 (2021): 51, 10.1186/s12902-021-00721-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
48. Zhang J., Luo H., Wu H., et al., “The Association Between Domestic Water Hardness and Kidney Stone Disease: A Prospective Cohort Study From the UK Biobank,” International Journal of Surgery 111, no. 2 (2025): 1957–1967, 10.1097/js9.0000000000002198. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials