Groundwater compliant with drinking water standards poses a “silent risk” of cumulative health hazards

Abstract

Groundwater is the primary source of drinking water for private well users. Traditional safety assessments and the classification of groundwater bodies (GWBs) typically rely on regulatory standards for individual contaminants, thereby overlooking mixture toxicity. This study aims to conduct a comprehensive health risk assessment (HRA) and evaluate existing regulatory standards within the framework of the “silent risk” phenomenon. Monitoring data spanning 2010, 2016, and 2022 for 174 GWBs in Poland were analysed. Non-carcinogenic (hazard index, HI) and carcinogenic risks (total cancer risk, TCR) were evaluated for both adults and children using deterministic and probabilistic (Monte Carlo simulation) approaches. The carcinogenic risk assessment conservatively assumed that all measured total chromium occurred in the form of Cr(VI).The results were compared against Polish, European Union (EU), and World Health Organization (WHO) standards. Although manganese exceeded regulatory limits in over 60% of monitoring points due to the geogenic background, it was not the primary driver of health risk. Instead, extreme hazards (HI > > 1) were driven by localized, anthropogenic spikes in nitrate and boron concentrations. Notably, children faced health risks more than twice as high as adults. Crucially, the study identified conservative, model-derived a “silent risk” phenomenon, where water fully compliant with regulatory standards for individual contaminants still posed a screening-level warning of unacceptable cumulative health risks (HI > 1) due to the additive effects of chemical mixtures. Furthermore, aggregate GWB classifications masked localized hotspots of critical health hazards. These findings indicate that, within the limits of our conservative modeling assumptions, selected contaminants and ingestion-only exposure framework for the selected contaminants, regulatory compliance does not inherently ensure chemical safety. This screening-level silent risk highlights how current frameworks fail to account for mixture toxicity and the heightened vulnerability of children. Consequently, HRA should be integrated into routine monitoring programs, and GWB status assessments must incorporate consumption-based metrics to effectively safeguard the health of private well users.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10653-026-03355-w.

Introduction

Groundwater constitutes the primary source of drinking water for a significant proportion of the population. It is of particular importance that millions of people worldwide, including in rural areas of Poland, rely heavily on raw groundwater from regional aquifer systems for daily consumption, often via decentralized or private wells. In such decentralized scenarios, comprehensive water quality monitoring is rarely performed systematically due to awareness gaps and economic barriersThis stems from a lack of awareness regarding the potential long-term health effects of consuming contaminated water, as well as from financial barriers (Malecki et al., 2017; Flanagan et al., 2015, 2016; Lade et al., 2024; Lavallee et al., 2023; Dotherow et al., 2024; Schultz et al., 2024). Furthermore, standard testing packages offered to well owners are usually limited to a few basic microbiological, physicochemical, and organoleptic parameters, entirely omitting the analysis of highly toxic trace elements and heavy metals, such as arsenic, boron, cadmium, or lead (Flanagan et al., 2016; Lade et al., 2024; Malecki et al., 2017). Meanwhile, the quality of these waters is increasingly threatened by the presence of inorganic contaminants of diverse origins, ranging from the natural geogenic background to intensive anthropogenic pressure (Akhtar et al., 2021; Arienzo et al., 2022; George et al., 2023; Lora-Ariza et al., 2024; Malecki et al., 2017; Mukherjee et al., 2024).

Traditionally, the safety of water intended for human consumption is evaluated by comparing the concentrations of individual contaminants with the applicable threshold values specified in legal regulations (e.g., national regulations, EU directives, WHO guidelines). This approach is based on the assumption that water meeting the standards for each parameter individually is safe for health. However, such a perspective is insufficient for health protection. It ignores the toxicity of chemical mixtures, where multiple contaminants occurring below permissible limits can accumulate in the body, generating unacceptable risks (Baken et al., 2018; Bolan et al., 2023; Kumar et al., 2019). Moreover, legal standards frequently do not account for the disproportionately higher sensitivity of specific demographic groups, particularly children. As a result, water officially deemed compliant with the law, or in the case of well supplies, unexamined for harmful components, may pose an unrecognized hazard (Kubicz et al., 2018; Raheja et al., 2024; Rao et al., 2021). To reliably estimate actual exposure and hazard, it is necessary to transition from simple comparative analyses to advanced, probabilistic health risk assessment (HRA) models that account for the temporal and spatial variability of concentrations.

Early studies on groundwater quality focused mainly on identifying exceedances of threshold values for individual contaminants. However, a growing body of scientific evidence indicates that compliance with applicable legal standards does not guarantee the absence of public health risks. The primary limitation of current regulatory frameworks, including the recast EU Directive, is the failure to account for the additive effect of chemical mixtures (Directive (EU) 2020/2184). As demonstrated in studies, the co-occurrence of contaminants at sub-threshold levels can result in a cumulative toxicological risk that exceeds the safe threshold. Particular attention in this context is directed toward the pediatric population. Due to a higher water intake-to-body weight ratio and immature detoxification mechanisms, children are characterized by a significantly higher sensitivity to heavy metal and nitrate exposure (Ayejoto et al., 2024; Alidadi et al., 2019; Valerko et al., 2022).

Another important aspect is the methodological evolution in risk assessment. The traditional deterministic approach, relying on mean or maximum values, often leads to the underestimation or overestimation of risk, ignoring environmental variability, which can be further modified by extreme events such as droughts (Kubicz et al., 2021). Therefore, probabilistic risk assessment models, such as the Monte Carlo simulation, are increasingly applied in hydrogeological studies, allowing for the inclusion of uncertainty and the determination of the probability of exceeding safe thresholds for the most sensitive segments of the population (Ali et al., 2021; Jamshidi et al., 2021; Kumar & Maurya, 2025).

It should also be noted that risk characterization is strongly dependent on the genesis of the contaminants. Geogenic contaminants, such as manganese or arsenic, are typically characterized by a stable, chronic occurrence linked to the geological background (Mukherjee et al., 2024; Podgorski et al., 2022). On the other hand, anthropogenic contaminants, particularly nitrates and boron derived from municipal wastewater and agriculture, tend to show extreme, localized, and episodic spikes in concentrations (Bonnière et al., 2024; Jia & Qian, 2025; Tawfeeq et al., 2024).

The current classification and chemical status assessment of groundwater bodies (GWBs) under Directive 2000/60/EC are based on environmental and ecological criteria, and do not account for the direct use of these waters for consumption purposes, including human drinking water. This gap implies that waters formally possessing a good chemical status from a hydrogeological perspective may pose a serious threat to public health from a toxicological standpoint, which justifies the necessity of developing integrated water safety indices (Kubicz et al., 2025). This study provides the systematic, macro-scale empirical validation of this regulatory gap using a multi-temporal national dataset. While deterministic and probabilistic health risk assessment (HRA) methods, including the cumulative Hazard Index (HI) and Monte Carlo simulations, are well-established screening-level tools, their application in national-scale groundwater management often lacks a critical regulatory context. Therefore, the novelty of this study lies not merely in the application of these mathematical tools to monitoring data, but in using them to decouple regulatory compliance from actual toxicological drivers. Specifically, this study distinguishes between geogenic exceedances and episodic anthropogenic hotspots. Furthermore, it evaluates how the spatial aggregation of data at the GWB level masks these localized risks, creating a false sense of security. The cumulative HI is a conservative screening tool based on simple dose addition. It does not prove that these chemicals interact synergistically in the body or guarantee adverse health effects, particularly since the analyzed contaminants often target different organs. However, even acknowledging these mathematical limitations and without calculating organ-specific risks, the concept of “silent risk” serves as a crucial early warning. It exposes a striking blind spot in our current regulations. Groundwater can perfectly pass single legal limit for individual contaminants, yet the combined HI still flags a potential health concern, leaving vulnerable groups like children unprotected.

The aim of this study is a multidimensional assessment of the spatio-temporal health risk (non-carcinogenic and carcinogenic) resulting from exposure to groundwater contaminants, accounting for differences in the sensitivity of individual population groups (adults vs. children). Statistical analyses were linked with deterministic and probabilistic Monte Carlo simulations across 174 GWBs. This approach allowed for an exploration of how “silent risk” manifests across a aquifer network. This study verifies whether current Polish, EU, and WHO legal standards genuinely protect human health. The focus was placed specifically on scenarios where these exploitable aquifers are used directly for consumption. In doing so, this work illustrates why traditional, single-contaminant compliance checks are insufficient.

Materials and methods

Study area and sampling

Groundwater samples for analysis were collected in 2010, 2016, and 2022 as part of the state environmental monitoring program. The reliability of the conducted health risk assessment was based on the implementation of a Quality Control and Quality Assurance (QC/QA) protocol at every stage of the research process. All groundwater samples were collected during the state environmental monitoring program’s diagnostic campaigns in the hydrological years 2010, 2016, and 2022, which correspond to the 1st (2010–2015), 2nd (2016–2021), and 3rd (2022–2027) River Basin Management Plan cycles under the Water Framework Directive, respectively. Field procedures were conducted in strict accordance with national standards (PN-ISO 5667-4:2003), which included intensive well purging until the full stabilization of physicochemical parameters and immediate sample preservation to maintain chemical integrity during transport. Laboratory analyses were performed by the Central Chemical Laboratory of the Polish Geological Institute—National Research Institute (PIG-PIB), an accredited facility (PCA) operating based on the PN-EN ISO/IEC 17025:2005 standard. To eliminate instrumental bias and verify measurement precision, the analytical workflow systematically incorporated the use of certified reference materials (CRM), laboratory blanks, and duplicate analyses. Particular emphasis was placed on data selection stringency; only results with an ion charge balance error (calculated according to PN-C-04638-02:1989) within the range of ± 10% were included in the final risk models. Furthermore, to handle unquantifiable samples with concentrations below the limit of detection (LOD), a conditional decision rule based on a 10% environmental detection frequency threshold was adopted (US Epa, 1989). For widely distributed contaminants with a detection frequency exceeding 5% per monitoring year (e.g., nitrates, boron, or arsenic), non-detect results were substituted with half of the detection limit (½ LOD) to maintain a conservative exposure scenario. Conversely, for highly localized or geogenically restricted trace elements detected in less than 5% of the samples (e.g., beryllium, chromium, or cadmium), these unquantifiable values were substituted with zero (0) to avoid artificial risk inflation driven by laboratory detection limits. The detailed annual QA/QC metrics, including exact LODs, missing data percentages, and specific substitution rates for all analyzed parameters, are fully documented in Table S1 (Supplementary Material).

The study encompassed an area of 174 GWBs in Poland, located within the Baltic Sea catchment area (99.7% of the study area), including the Vistula (55.7%) and Odra (33.9%) river basins, as well as the Przymorze, Pregolya, and Neman rivers. Furthermore, the analysis included the Black Sea catchment area (0.19% of the area) with the Danube and Dniester river basins, and the North Sea catchment area (0.8% of the area) along with the Elbe river basin. Due to the expansion of the monitoring network in subsequent monitoring cycles, the number of sampling points utilized in the analyses varied, amounting to 781 points in 2010, 1290 points in 2016, and 1406 points in 2022. The detailed number of monitoring points in 174 GWB for each analyzed year is shown in the Supplementary materials (Table S2). Although the data originate from the state monitoring network, these sampling points tap the same major exploitable aquifers that are commonly utilized by private well users. Private wells vary in depth, construction, and local protection; therefore, this national dataset is not used to predict exact water quality in specific domestic wells. Instead, it serves as a robust regional baseline proxy for raw groundwater quality. While it may not capture micro-scale variations introduced by shallow household well construction, the chemical status observed at these regional monitoring points provides a reliable macro-scale estimation of the baseline exposure for unmonitored users of private intakes.

Statistical analysis

To verify whether the observed differences in the concentrations of the analysed chemical parameters among the three monitoring periods (2010, 2016, and 2022) were statistically significant, the non-parametric Kruskal–Wallis test was applied. The choice of this test was dictated by the specific nature of environmental data. The concentration distributions of contaminants in groundwater did not meet the assumptions of normality, and the sample sizes (number of monitoring points) varied across the respective study years. The level of statistical significance was set at p < 0.05. The results of this test allowed for the differentiation of actual changes in water chemistry from fluctuations resulting solely from the natural dispersion of data and the temporal expansion of the monitoring network.

Regulatory framework and standards comparison

The obtained contaminant concentrations were compared with the applicable threshold values specified in Polish (PL) and European Union (EU) regulations, as well as WHO guidelines. A list of these values, along with the oral reference doses (RfD) used in the risk assessment, is presented in Table 1. It should be noted that for certain parameters (e.g., chromium and lead in the EU standard), target values are provided, and instances where no threshold value is established in a given regulation are indicated with a dash (-). The data sources for determining the specific values included the Integrated Risk Information System of the U.S. Environmental Protection Agency, Edition of the Drinking Water Standards and Health Advisories Tables (Report No. EPA 822-F-18-001) for the oral reference doses (RfD) and cancer slope factors (SF); Guidelines for drinking-water quality: Fourth edition incorporating the first and second addenda, World Health Organization, Geneva, 2022 for the WHO Standard (WHO); Directive (EU) 2020/2184 of the European Parliament and of the Council of 16 December 2020 on the quality of water intended for human consumption (recast) for the European Union Standard (EU); and the Regulation of the Minister of Health of 7 December 2017 on the quality of water intended for human consumption (Journal of Laws 2017, item 2294) for the Polish Standard (PL). The SF values for As and Cr were obtained from the US EPA IRIS database. Due to the lack of current oral SF values for lead in the US EPA IRIS database, the value developed by the California Office of Environmental Health Hazard Assessment (OEHHA, 2009) was used for this element. Since the monitoring program does not include chromium speciation, a conservative approach (worst-case scenario) was assumed, positing that the total measured chromium occurs in the form of Cr(VI). While the majority of threshold values across PL, EU, and WHO frameworks are strictly health-based, it is important to note that regulatory standards classify certain substances as indicator parameters. Manganese serves as the prime example of this distinction. In both EU and PL regulations, the strict 0.05 mg/L limit for Mn functions primarily as an aesthetic and operational indicator. This threshold is intended to prevent consumer rejection due to laundry staining, taste alterations, and pipe encrustation, rather than to prevent direct toxicological harm. In contrast, the toxicological Reference Dose (RfD) for Mn used in this HRA is relatively high (0.14 mg/kg/day). When back-calculated using standard exposure factors, this RfD translates to a theoretical health-based limit of approximately 4.9 mg/L for adults. Consequently, while Mn frequently exceeds the 0.05 mg/l regulatory indicator limit due to the natural geogenic background, its actual toxicological contribution to the Hazard Index (HI) remains marginal. This regulatory distinction explains why a high frequency of legal non-compliance for Mn does not automatically equate to a high health risk in HRA modeling.

Table 1.

Comprehensive summary of toxicological parameters and water quality standards

Compound	WHO (mg/l)	European Union (mg/l)	Poland (mg/l)	RfD (mg/kg day)	SF (mg/kg day)
Antimony (Sb)	0.02	0.01	0.005	0.0004	–
Arsenic (As)	0.01	0.01	0.01	0.0003	1.5
Nitrates (NO3)	50	50	50	1.6	–
Nitrites (NO2)	3	0.5	0.5	0.1	–
Barium (Ba)	1.3	–	0.7	0.2	–
Beryllium (Be)	–	–	–	0.002	–
Boron (B)	2.4	1.5	1	0.2	–
Chromium (total assumed)	0.05	0.025	0.05	0.003	0.5
Zinc (Zn)	–	–	–	0.3	–
Cadmium (Cd)	0.003	0.005	0.005	0.0005	–
Manganese (Mn)	0.08	0.05	0.05	0.14	–
Nickel (Ni)	0.07	0.02	0.02	0.02	–
Lead (Pb)	0.01	0.005	0.01	–	0.0085
Selenium (Se)	0.04	0.02	0.01	0.005	–
Strontium (Sr)	–	–	–	0.6	–
Uranium (U)	0.03	0.03	–	0.003	–

Health risk assessment

The health risk assessment resulting from exposure to groundwater contaminants was conducted in accordance with the guidelines of the United States Environmental Protection Agency (US Epa, 2001, 2010, 2011). In the exposure assessment model, only the dose of substances entering the body through water ingestion was considered. Although dermal absorption and inhalation during showering represent potential exposure pathways for groundwater contaminants, this study focuses exclusively on the ingestion pathway. Ingestion is widely recognized as the primary and most conservative driver of systemic toxicity for the specific target contaminants analysed here, such as nitrates and boron. While omitting the dermal pathway may slightly underestimate the total cumulative risk, it ensures a highly focused evaluation of direct drinking water consumption hazards. Other potential exposure pathways, such as dermal contact or inhalation, were excluded. Therefore, the estimated risk pertains solely to water consumption. The calculations were performed separately for two demographic groups: adults and children. The Chronic Daily Intake (CDI) for the ingestion pathway was calculated using the following equation:

$$CDI=\frac{C·IR·EF·ED}{Bw·AT}\text{mg/kg d}$$ 1

where

CDI is the chronic daily intake (mg/kg d),

C is the concentration of the contaminant in water (mg/l),

IR is the water ingestion rate (assumed to be 2.0 l/day for adults and 1.0 l/day for children, in accordance with US EPA guidelines),

EF is the exposure frequency (365 days/year),

ED is the exposure duration (assumed to be 30 years for adults and 6 years for children),

BW is the body weight (assumed to be 70 kg for adults and 15 kg for children),

AT is the averaging time (for non-carcinogenic effects AT = ED · 365, AT = ED 365 · days for carcinogenic effects AT = 70 years · 365 days).

The risk of adverse non-carcinogenic health effects was evaluated using the Hazard Quotient (HQ) and the Hazard Index (HI):

$$\text{HQ}=\text{CDI}/\text{RfD}$$ 2

where RfD is the oral reference dose (Table 1).

The cumulative non-carcinogenic risk from simultaneous exposure to multiple contaminants was expressed as the HI, which is the sum of the individual HQ values:

$$\text{HI}=\mathrm{\Sigma }{\text{HQ}}_{\text{i}}$$ 3

where

HQ_i–hazard quotient for the i^th substance.

An HI value < 1.0 indicates that the exposure does not pose a significant risk of adverse health effects, whereas an HI value > 1.0 indicates the probability of toxic effects occurring.The carcinogenic risk (Total Cancer Risk, TCR) was estimated as the probability of developing cancer as a result of lifetime exposure to carcinogenic substances (As, Cr, and Pb were included in the analysis). It was calculated by multiplying the chronic daily intake (CDI) by the cancer slope factor (SF), and subsequently summed for all carcinogenic compounds:

$$\text{CR}=\text{CDI}·\text{SF}$$ 4

$$\text{TCR}=\mathrm{\Sigma }{\text{CR}}_{\text{i}}$$ 5

According to US EPA guidelines, the acceptable carcinogenic risk range 10^–6–10^–4 (i.e., from 1 case per million to 1 case per 10,000 people). A TCR value exceeding 10^–4 is considered an unacceptably high health risk.

Due to the absence of an established oral cancer slope factor (SF) for lead in the US EPA IRIS database which currently classifies the oral route as inadequate for quantifying carcinogenicity the threshold developed by the California Office of Environmental Health Hazard Assessment (OEHHA, 2009) was adopted. Following the methodology established in health risk literature (Hossain et al., 2018), this specific oral slope factor (0.0085 mg/kg/day), derived from epidemiological data linking oral lead exposure to internal carcinomas, was utilized as a scientifically justified surrogate. Its application prevents the systematic underestimation of lead-induced carcinogenic effects in the total cancer risk (TCR) models, ensuring a protective screening approach within this national-scale groundwater assessment.

TCR value exceeding 10^–4 is considered an unacceptably high health risk. Because the state monitoring program does not include chromium speciation, the baseline assessment (Scenario I) adopted a highly conservative approach, assuming that 100% of the measured total chromium occurs in the carcinogenic Cr(VI) form. To evaluate the robustness of the carcinogenic risk assessment and address the potential overestimation of Cr(VI) fractions, a sensitivity analysis was conducted. Two alternative scenarios were modeled: Scenario II, assuming Cr(VI) constitutes 10% of total chromium (typical for well-oxygenated shallow groundwaters), and Scenario III, assuming a 1% Cr(VI) fraction (characteristic of reduced, deep aquifer environments where Cr(VI) is reduced to Cr(III)).

Probabilistic analysis (Monte Carlo)

To account for the uncertainty and variability of the input parameters (primarily the variability of contaminant concentrations), a Monte Carlo simulation was applied in addition to the deterministic assessment. The analysis was conducted for 10,000 iterations, calculating the probability distributions for the HI and TCR indices. The results were presented as the median value (representing a typical consumer) and the 95th percentile (representing the worst-case scenario or the most sensitive segment of the population). This approach is established and accepted in scientific research (Ali et al., 2023; Guron et al., 2024; Imtiyaz & Rather, 2025). In the simulation, only contaminant concentrations were treated as variables. We applied a non-parametric empirical distribution. The algorithm resampled directly from the raw monitoring datasets for each year. All human exposure factors (IR, BW, ED, EF, AT) remained fixed at US EPA point estimates. This isolates the output variance. It reflects true spatial groundwater heterogeneity rather than human behavioral variability. The procedure was scripted in Python using the NumPy library.

Results

Descriptive statistics and standards comparison

A comparison of the analytical results with the applicable national legal standards and international guidelines revealed significant exceedances, among which manganese (Mn) was definitively the dominant regulatory and aesthetic compliance issue, rather than a primary toxicological driver (Fig. 1). In 2010, approximately 59.9% of the monitoring points (468 out of 781) exceeded the Polish (PL) and European Union (EU) standards for Mn. The WHO guidelines were slightly less stringent; however, exceedances were still recorded at a very high level of 52.0% of the points (406)., This issue intensified over the years. In 2016, the percentage of PL and EU standard exceedances increased to 64.9% (837 out of 1290 points), and in 2022, it reached 65.0% (914 out of 1406 points). The mean Mn concentration remained at a constantly high level (approx. 0.20 mg/L) during the analyzed years, with maximum values reaching over 5.0 mg/L (5.09 mg/L in 2010, 5.71 mg/L in 2016, and 5.01 mg/L in 2022), which repeatedly exceeded the permissible limits. In addition to manganese, the analysis revealed systematic, albeit much smaller-scale, exceedances for nitrates (NO₃⁻)) and arsenic (As). The percentage of exceedances for nitrates (NO₃⁻) under the PL, EU, and WHO standards (which are congruent for this parameter) grew during the study period from 4.1% in 2010 (32 points), through 4.5% in 2016 (58 points), to 4.8% in 2022 (67 points). Notably, very high maximum values were observed (e.g., 280 mg/L in 2016 against the standard of 50 mg/L), indicating strong point-source anthropogenic contamination. Exceedances of arsenic (As) concentrations remained at a similar level throughout the entire analyzed period, amounting to approx. 2.3% in 2010, 3.0% in 2016, and 2.7% in 2022. All three standards (PL, EU, WHO) showed an identical number of exceedances here. The maximum values were very high, ranging from 0.108 mg/L to 0.343 mg/L.The analysis also revealed significant differences in the classification of exceedances depending on the adopted standard (PL, EU, WHO) for several other parameters. In 2010, a clear discrepancy was noted between the standards for cadmium (Cd). There were 2.0% exceedances according to PL and EU standards (16 points) and as much as 4.0% according to WHO guidelines (31 points), which results from the more restrictive WHO standard. In subsequent years, the cadmium problem was almost completely resolved, as the percentage of exceedances under all standards fell below 0.2% in 2016 and 2022. In the case of boron (B), discrepancies between the standards were visible throughout the analyzed period. In 2010, PL standards were exceeded in 2.0% of points (16), EU standards in 1.3% (10), and WHO guidelines in 1.0% (8). A similar trend continued in 2016 and 2022, with a predominance of PL standard exceedances over the EU and WHO standards. Historical legacies or the geological background cause boron concentrations to be very high at certain points (the maximum value in 2010 reached as much as 132.79 mg/L). The analysis of nickel (Ni) and nitrite NO-2) concentrations showed that for both parameters, Polish and EU regulations exhibit more frequent exceedances than the WHO guidelines. For instance, in 2022, the PL and EU standards for Ni were exceeded in 1.4% of points (19), whereas the WHO standard was exceeded in only 0.1% (2 points). Analogously, for NO-2 in 2022, this percentage was 0.9% (12 points) according to PL and EU standards, and a mere 0.1% (2 points) according to the WHO (Fig. 2).For the remaining analyzed parameters, such as antimony (Sb), barium (Ba), beryllium (Be), chromium (Cr), zinc (Zn), selenium (Se), and uranium (U), the percentage of exceedances was marginal (typically below 0.3% of points) or not recorded at all. These phenomena were exclusively of a point-source nature and did not pose a general threat to groundwater quality on a regional scale. However, certain discrepancies between the standards for point-source contaminants should be noted. In 2016, three exceedances were recorded for lead (Pb) exclusively under the EU standard, while for uranium (U) in the same year, three exceedances each were recorded according to the EU and WHO standards.

Fig. 1.

Location of measurement points in Poland within the Baltic Sea catchment area in relation to Groundwater Bodies (GWBs)

Fig. 2.

Discrepancies in the classification of exceedances of standards over time (selected parameters: Mn, As, NO3, B, Cd, Ni)

Significance testing

To verify whether the differences in parameter concentrations observed among the years 2010, 2016, and 2022 were statistically significant, a non-parametric Kruskal–Wallis test was conducted. The results of the test (at a significance level of p < 0.05) confirmed that for the majority of the analysed indicators, the temporal changes were not accidental, although they also revealed several important anomalies. The most surprising result is the lack of statistical significance for manganese (Mn) (p = 0.127). The increasing percentage of Mn exceedances results from the expansion of the monitoring network. New monitoring points were likely established in zones with an elevated Mn background. This phenomenon is well demonstrated by the statistics. Although in the descriptive analysis manganese exhibited the highest percentage of regulatory exceedances (increasing from 59.9 to 65.0%) and constituted the primary water quality problem, its concentration distribution did not undergo a significant change over the 12-year period. This indicates that manganese contamination is chronic and stable, and the observed percentage fluctuations result solely from the natural dispersion of data and the increase in the number of monitoring points (N), rather than from an actual deterioration or improvement in water quality regarding this parameter.

Conversely, changes in concentrations for other key parameters identified in the first step proved to be statistically significant. For nitrates (NO-3) (p = 0.037) and arsenic (As) (p = 0.042), significant differences were observed among the years, confirming that the recorded fluctuations in the percentage of exceedances reflect actual changes in the aquatic environment. Exceptionally high statistical significance (p ≈ 0.000) was demonstrated for cadmium (Cd) and nickel (Ni). In the case of cadmium, this is fully consistent with previous findings. The sharp decline in the percentage of exceedances from 2.0% (according to PL and EU standards) in 2010 to a mere 0.08% in 2022 was indisputably confirmed statistically as a clear improvement in water quality. A similar situation is observed for nickel, where concentration distributions differed significantly across the studied years. Highly distinct temporal changes (p < 0.001) were also recorded for chromium (Cr), zinc (Zn), and lead (Pb). Although these parameters rarely exceeded regulatory standards, their background concentrations underwent significant transformations over the years. On the other hand, alongside manganese, the group of parameters for which no statistically significant changes in concentrations were demonstrated over the analyzed period (p > 0.05) consists of boron (B), uranium (U), beryllium (Be), and selenium (Se). In the case of boron, despite the previously observed differences in the classification of exceedances among the PL, EU, and WHO standards, the concentrations of this element remained at a stable level throughout the entire period. Similarly stable and unchanging over time were the concentrations of selenium and uranium, for which exceedance issues were exclusively point-source and incidental. Statistically significant changes in concentrations that did not translate into compliance issues (due to low absolute concentrations) were recorded for antimony (Sb) (p = 0.001) and barium (Ba) (p = 0.017).

In summary, the Kruskal–Wallis test demonstrates that the soil-groundwater environment underwent significant transformations primarily with respect to heavy metals (Cd, Ni, Pb, Cr, Zn) and selected anions (NO₃, NO₂). The greatest water quality problem is associated with the presence of manganese. It exhibits the characteristics of a multi-year phenomenon, without signs of clear positive or negative trends.

Deterministic and probabilistic risk

In developing the next stage of the analysis, the health risk assessment results (HI and TCR indices) were linked to the previous findings regarding exceedances of legal standards and guidelines. Data analysis disaggregated by groundwater bodies (GWBs) and age groups revealed critical, unexpected patterns that stand in contrast to a simple comparative analysis against regulatory standards. The health risk assessment demonstrated that the consumption of groundwater from the analysed units (GWBs) carries a certain level of hazard, with this risk being spatially differentiated and dependent on the age group. A significantly greater vulnerability of the pediatric population was clearly pronounced. In all analyzed monitoring points, the HI values for children were, on average, more than twice as high as those for adults, which stems from lower body weight and a higher relative water intake.

Non-carcinogenic risk (HI) exceeded the safe level (HI > 1.0) at numerous monitoring points, and in extreme cases, reached alarming values. For example, in the GWB 168 unit, the HI was recorded at 32.4 for adults and as high as 75.7 for children, while in GWB 166, the values were 19.6 and 45.9, respectively. This implies that in these locations, the risk of adverse health effects is very high, and suspending the use of these waters for consumption purposes is absolutely required. Carcinogenic risk (TCR), accepted as safe at a level below 1.10⁻⁴ (1 case per 10,000 people), remained within the range of 1.10⁻⁵–1.10⁻⁴ at most points. However, locations with unacceptably high carcinogenic risk exceeding the 1.10⁻⁴ threshold were identified, reaching values on the order of 9.10⁻³ or 6.10⁻³ in extreme cases (e.g., GWB 22, GWB 168). A key conclusion from this analysis is the lack of a direct correlation between legal standard exceedances (PL, EU, WHO) and actual health risk, which is particularly evident against the background of previous findings. In the preceding stage, it was demonstrated that manganese (Mn) was the parameter with the highest percentage of regulatory exceedances (approx. 60–65%). Meanwhile, the risk analysis reveals that at many points where standard exceedances were recorded, the HI index was lower than 1.0 (e.g., a point in GWB 20 showed an HI for adults = 0.31 and an HI for children = 0.72). This indicates that although Mn or other metal concentrations exceed stringent legal standards, they do not always generate an immediate toxicological risk in terms of the HI index. Conversely, the highest non-carcinogenic risk values (HI ≫ 1.0) were registered at points where no standard exceedances were recorded. This fact directly relates to the data from the first part of the analysis, where isolated but extremely high maximum concentrations were found for nitrates (NO₃) and boron (B) (e.g., 280 mg/L for NO₃ and 132 mg/L for B). It is these parameters, at points with the poorest water quality, that constitute the dominant factor causing high HI values above the safe level, particularly for the pediatric population.

The comparison of PL, EU, and WHO standard exceedances with TCR values shows that carcinogenic risk (driven primarily by metals such as As, Cr, Pb, and Cd) can be significant (TCR > 10^–4) even at points where no exceedances of statutory standards were recorded. This indicates that in carcinogenic risk assessment, the reference slope factors adopted in the models are more sensitive to low concentrations of specific contaminants than the threshold values established in regulatory standards.

In summary, the health risk assessment proved that exclusive reliance on comparing concentrations with legal standards can lead to erroneous conclusions regarding the actual health hazard. Manganese, which exhibits the most frequent exceedances, is not the primary driver of risk. The most severe threats to public health (HI > 1.0 and elevated TCR) are associated with localized, extreme contamination by chemical compounds (predominantly from regions where nitrate and boron concentrations reach maximum values), with the pediatric population representing a particularly vulnerable group.

The Monte Carlo simulation results demonstrated a clear discrepancy between the simulation utilizing the median values and the risk under the worst-case scenario according to the 95th percentile (Fig. 3). Non-carcinogenic risk (HI) at median values for adults remained at a relatively low, safe level, ranging from 0.267 (in 2016) to 0.298 (in 2010). For the pediatric population, the median HI was more than twice as high, oscillating around 0.70–0.79. This implies that for a typical consumer of water from the analyzed intakes, the risk of toxic effects does not exceed the permissible threshold (HI = 1.0). However, the 95th percentile analysis reveals the scale of the threat to the most sensitive segment of the population. The 95th percentile for adults exceeded a value of 1.6 in all analyzed years, and for children, it reached alarming values, ranging from 4.20 (in 2016) to 4.56 (in 2010). This result directly confirms the conclusions of the deterministic analysis. It is precisely the extreme concentrations of contaminants (such as nitrates or boron) at selected monitoring points that account for the unacceptably high health risk for children, whereas the average water quality does not raise serious toxicological concerns. Carcinogenic risk (TCR) exhibited a different pattern compared to HI. The median TCR throughout the entire analyzed period was equal to 2.76·10⁻⁵ for adults and 1.29·10⁻⁵children, indicating that over 50% of the analyzed monitoring points showed no measurable carcinogenic risk from the investigated elements. Conversely, the 95th percentile for adults in 2016 and 2022 slightly exceeded, or equaled, the safe threshold, while for children and adults in 2010 it remained below this limit. In contrast to the non-carcinogenic risk, the TCR at the 95th percentile level was higher for adults (ranging from 1.01·10⁻⁴ to 1.19·10⁻⁴) than for children (ranging from 4.72·10⁻⁵ do 5.57·10⁻⁵). This stems from the fact that in carcinogenic risk assessment models, the dose is often accumulated over time (a longer anticipated exposure duration over a lifetime for adults), which makes metals with carcinogenic effects (e.g., As, Cd, Cr, Pb) pose a relatively greater threat to the adult population. The temporal changes identified in the Monte Carlo simulation are fully consistent with the results of the Kruskal–Wallis test. For both HI and TCR, the year 2010 was characterized by the highest 95th percentile values. The decline in carcinogenic risk is a direct reflection of the statistically proven significant improvement in water quality regarding heavy metals (specifically cadmium). The high, although decreasing, non-carcinogenic risk at the 95th percentile level confirms the persistent issue associated with high concentrations of nitrogen compounds and boron.

Fig. 3.

The gap between the median and the 95th percentile of health risk (HI) in the Monte Carlo simulation

To evaluate the robustness of the carcinogenic risk assessment and address the potential overestimation of Cr(VI) fractions, a sensitivity analysis was conducted based on the reviewer’s recommendation. Because the state monitoring program does not include chromium speciation, the baseline assessment (Scenario I) adopted a highly conservative approach, assuming that 100% of the measured total chromium occurs in the carcinogenic Cr(VI) form. Two alternative scenarios were modeled: Scenario II, assuming Cr(VI) constitutes 10% of total chromium (typical for well-oxygenated shallow groundwaters), and Scenario III, assuming a 1% Cr(VI) fraction (characteristic of reduced, deep aquifer environments where Cr(VI) is reduced to Cr(III)).

The results of this sensitivity analysis demonstrated that the carcinogenic risk is overwhelmingly driven by arsenic, rendering the overall TCR highly insensitive to chromium speciation assumptions. For adults, the 95th percentile TCR under Scenario I (100% Cr(VI)) ranged from 1.01·10⁻⁴ to 1.19 × 10⁻⁴ across the analyzed years. Under the more realistic Scenario II (10% Cr(VI)), the 95th percentile TCR decreased only marginally, ranging from 9.28·10⁻⁵ to 1.11·10⁻⁴. Even under the extremely conservative Scenario III (1% Cr(VI)), the 95th percentile values remained virtually unchanged (9.20·10⁻⁵–1.10 × 10⁻⁴). A similar pattern was observed for children, where the 95th percentile TCR shifted only slightly from 4.72·10⁻⁵ to 5.57·10⁻⁵ (Scenario I) to 4.29 × 10⁻⁵ to 5.15·10⁻⁵(Scenario III).

This negligible impact of chromium on the total cancer risk is mathematically explained by the dominance of arsenic. At the 95th percentile, the cancer risk attributable solely to arsenic accounts for approximately 91% to 92% of the total TCR in both adults and children. Consequently, even a 100-fold reduction in the assumed Cr(VI) fraction (from Scenario I to Scenario III) results in a maximum TCR reduction of less than 8%. This analysis confirms that the identified exceedances of the acceptable carcinogenic risk threshold (10⁻⁴) for adults in 2016 and 2022 are fundamentally attributed to geogenic arsenic exposure, not chromium. Therefore, while Scenario I represents a worst-case boundary condition, the conclusion regarding unacceptable localized carcinogenic risks remains robust regardless of the actual, unknown Cr(VI) speciation in these groundwater systems.In summary, the probabilistic approach demonstrates that the problem of health risk related to groundwater quality is not widespread, but rather localized in nature.

Temporal and spatial trend analysis (GWB)

Aggregating the risk assessment data to the level of groundwater bodies (GWBs) revealed spatial variations in the potential health hazard. It was confirmed that the non-carcinogenic risk (HI) problem is localized in nature, with a clear dominance of the hazard toward the pediatric population. The majority of GWBs are characterized by a relatively safe level of health risk (HI_Median < 1.0). However, the analysis of the worst-case scenario based on the 95th percentile of HI values allowed for the identification of areas with critical hazard levels, where the HI for children exceeds the safe threshold multiple times. In 2010, the GWB 168 unit recorded the absolute highest values in the entire dataset (HI_P95 reached 94.75 for children and 38.83 for adults). A similarly high level was recorded in the GWB 125 unit in 2016 (HI_P95 for children = 43.79, adults = 18.95). This result is particularly intriguing given that in 2010, the P-95 value for children in this unit was only 2.24, and in 2022, it fell again to a level of 1.38. This drastic, sharp increase in risk in 2016, followed by an equally rapid decline, stands in sharp contrast to the stable nature of manganese contamination and strongly suggests an episodic, point-source discharge of anthropogenic contaminants (nitrates) into the soil-groundwater environment during that period. Additionally, high values were recorded in GWB 102 in 2010 (HI_P95 for children = 13.76).

Such high HI values confirm the previous findings that they are driven by extreme concentrations of nitrates and boron, rather than by the widespread occurrence of manganese. Certain GWB units are characterized by persistently high and simultaneously stable risk. An example is GWB 43, where the HI_P95 for children remained at a very high but stable level throughout the entire period: 12.50 (2010), 13.31 (2016), and 11.65 (2022). A similar phenomenon is observed in GWB 73 (HI_P95 for children within the range of 6.0–6.5).The analysis of temporal changes (2010 vs. 2016 vs. 2022) showed that the health risk trend depends on the location. GWBs were identified where effective measures were taken or the source of contamination was eliminated. As illustrated in Fig. 4, the most spectacular decline in risk was recorded in the GWB 168 unit, where the HI_P95 for children dropped from 94.75 (2010) to a mere 0.87 (2022). Toxicologically, this risk was initially driven by an extreme concentration of arsenic (Max Ratio = 1143 in 2010) and decreased by orders of magnitude to a completely safe level in subsequent years. A clear improvement was also recorded in the GWB 125 unit, which experienced a drastic, sharp increase in risk in 2016 (HI_P95 for children = 43.79; Max Ratio = 620, with As as the driver), followed by an equally rapid decline in 2022. Similar downward trends were identified in the GWB 38 and GWB 109 units. Unfortunately, areas with a deteriorating situation were also observed. In GWB 21, the risk for children (P-95) increased drastically from 5.05 (2010) to 16.48 (2022). A similarly concerning upward trend was recorded in GWB 32 (an increase in P-95 from 1.72 to nearly 10.0) and GWB 24, indicating the emergence of new sources of contamination in these areas. These dynamic, spike-like changes in risk values over time-both rapid increases and decreases-stand in clear contrast to the stable nature of geogenic contamination (by manganese) and strongly suggest that episodic discharges of anthropogenic contaminants into the soil-groundwater environment shape the spatial and temporal pattern of public health risk.

Fig. 4.

Temporal variability of maximum health risk (Max Ratio) in selected groundwater bodies (GWBs) driven by anthropogenic point-source pollution

A comparison of the overall Monte Carlo simulation results (where the general P95 for children was approx. 4.2–4.5) with the disaggregated analysis per GWB proves that aggregate statistics do not capture the full extent of the problem. The overall P95 value is the result of averaging safe units with extremely contaminated ones. Spatial aggregation at the GWB scale inevitably introduces statistical smoothing, risking the overgeneralization of isolated extremes. To resolve this, we assessed the temporal persistence of high-risk classifications. The data reveal that extreme 95th percentile HI values are predominantly driven by episodic, single-point anomalies rather than pervasive aquifer contamination. For instance, the drastic risk spike in GWB 125 in 2016 (HI_P95 = 43.79 for children), contrasted with safe baselines in 2010 (2.24) and 2022 (1.38), hydrogeologically indicates a localized, transient contaminant plume rather than a regional failure. In contrast, persistent, multi-point hazards were rare, exemplified only by units like GWB 43, where HI_P95 remained consistently elevated (> 11.0) across all cycles. Consequently, aggregate GWB statistics mask localized hotspots. The national median reflects the true regional baseline, while extreme percentiles act merely as spatial flags for targeted, point-scale investigations.

The temporal changes identified in the Monte Carlo simulation are fully consistent with the results of the Kruskal–Wallis test. For both HI and TCR, the year 2010 was characterized by the highest 95th percentile values. The decline in carcinogenic risk is a direct reflection of the statistically proven significant improvement in water quality regarding heavy metals (specifically cadmium). The high, although decreasing, non-carcinogenic risk at the 95th percentile level confirms the persistent issue associated with high concentrations of nitrogen compounds and boron.

In conclusion, the probabilistic approach demonstrates that the problem of health risk related to groundwater quality is not widespread, but rather localized in nature.

Standard exceedance versus health risk

The final stage of the study involved cross-referencing the compliance of water quality assessments under legal regulations (PL, EU, and WHO standards) with the calculated health risk (HI). This analysis revealed a phenomenon we term the “silent risk” scenario. Crucially, this must be defined strictly as a screening-level warning rather than confirmed cumulative toxicity. It identifies cases where water officially meets all legal requirements for individual contaminants, yet the calculated non-carcinogenic hazard index (HI) exceeds the safe threshold (HI > 1.0). Because the HI is a conservative, additive screening metric, this threshold breach results from combining sub-threshold exposures and applying child-specific intake rates, rather than proving synergistic toxicological interactions or clinical harm. The visualization of this phenomenon presented in Fig. 5 clearly demonstrates that exceedances of legal standards for manganese (points to the right of the Mn standard line) do not translate into high Max Ratio values. Simultaneously, arsenic and boron generate extreme toxicological risk at concentrations on the order of 0.05–0.1 mg/L for As and 2–5 mg/L for B, which frequently oscillate around or only slightly exceed the legally permissible threshold values. This directly confirms that legislation based on threshold values for individual substances fails to account for mixture toxicity or the specific sensitivity of children.

Fig. 5.

The “silent risk” phenomenon, relationship between the concentration of dominant risk drivers and their toxicological contribution (Max Ratio) for example compared with PL limitations

In 2010, within GWB 84, water at as many as 7 monitoring points met the regulatory standards while simultaneously generating an unacceptable health risk (HI > 1.0). In 2016, a similar concentration of “silent risk” was recorded in units such as GWB 130 (5 points according to the EU standard) and GWB 110 (6 points according to the WHO guideline). The situation did not improve in 2022. In GWB 84, the number of such points increased to 11, and in GWB 90, 5 such cases were recorded. A comparison of the results across different legal standards shows minor yet notable differences in detecting “silent risk”. The PL and EU standards present an almost identical pattern, which is the result of the extensive harmonization of Polish law with EU legislation. The WHO guidelines, in certain cases, exhibit greater sensitivity to silent risk. For instance, in 2016, 4 points with such a discrepancy were recorded in GWB 110 under the PL and EU standards (i.e., points where water complied with legal standards but the HI risk exceeded 1.0), whereas as many as 6 points were identified under the WHO standard. This stems from the fact that certain WHO threshold values are more restrictive (as in the case of cadmium). Consequently, water that does not meet the WHO standard is immediately classified into the high risk category, thereby eliminating the phenomenon of “silent risk” for that specific monitoring point. It must be noted, however, that even more stringent WHO standards cannot prevent a scenario where multiple different contaminants—each individually within the norm—accumulate in the body and collectively exceed the safe risk threshold (HI > 1.0). The occurrence of “silent risk” provides direct confirmation of the findings from the previous steps and can be explained by two primary mechanisms. Generally accepted legal guidelines define threshold values for each parameter individually. Conversely, the HI index aggregates the risk from multiple different contaminants. Water may contain concentrations of arsenic, boron, nitrates, and manganese at levels representing, for example, 60–80% of their respective legal standards. From a legal standpoint, such water is suitable for consumption. However, from a toxicological perspective, the cumulative body burden (HI) exerted by these four substances simultaneously exceeds 1.0, thereby indicating a health hazard. Furthermore, legal standards are frequently calibrated based on an adult human with a standard body weight and metabolism. As demonstrated previously, the risk for children is twice as high as that for adults. Water that is legally safe for an adult (HI-adult < 1.0) frequently generates an HI > 1.0 for children due to their lower body weight and higher relative water intake. The identification of the “silent risk” phenomenon constitutes the most critical argument that relying exclusively on regulatory criteria to evaluate groundwater quality is insufficient for the protection of public health. Legislation fails to account for the cumulative effect of multiple contaminants present in water at sub-threshold concentrations, as well as the specific sensitivity of children. Demonstrating the discrepancies between legal compliance and actual health risk (HI > 1.0) proves that health risk assessment should constitute an additional, standard component of monitoring programs for water intended for human consumption, particularly in areas with complex chemical characteristics, such as units GWB 84, GWB 90, or GWB 110.

Discussion

This study provides evidence that the traditional approach to evaluating groundwater safety, which relies on comparing concentrations with threshold values for individual substances, is insufficient for public health protection. The primary finding of this work is the identification and characterization of the “silent risk” phenomenon, wherein water officially deemed compliant with legal standards generates unacceptable cumulative health risks (HI > 1.0). It should be noted that the standard Hazard Index (HI) approach assumes additivity of toxic effects, which serves as a highly conservative Screening-level tool. Since the analysed contaminants (e.g., nitrates, boron, and manganese) possess different toxicological mechanisms and target organs, the calculated HI reflects a worst-case cumulative exposure scenario rather than a strict biological synergy. The standard Hazard Index (HI) approach used in this study assumes that toxic effects are strictly additive, which provides a highly conservative screening-level assessment. Because the analysed contaminants, such as nitrates, boron, and manganese, act through different toxicological mechanisms and target different organs, the calculated total HI represents a worst-case cumulative exposure scenario rather than definitive biological synergy. Defining target-organ-specific HIs was beyond the scope of this national scale screening. This is primarily because standard US EPA guidelines do not establish organ-specific reference doses for all investigated parameters, which would make a unified organ-specific aggregation unreliable across such a large dataset. Consequently, the total HI reported here should be interpreted as a conservative screening benchmark designed to flag cumulative environmental stress, rather than direct evidence of localized organ toxicity or confirmed chemical interactions. Nevertheless, this approach is crucial for public health protection, as it signals potential multi-contaminant stress that traditional single-substance thresholds fail to detect., This issue is particularly concerning in the context of the widespread use of private water supplies (wells), where standard testing packages rarely include a toxicological analysis of contaminant mixtures. The “silent risk” phenomenon arises from the additive effect of accumulating multiple contaminants, each occurring below its permissible limit, yet their combined interaction within the body exceeds the safe toxicological threshold. This directly challenges current regulatory frameworks, including the recast EU Directive, which still fails to account for mixture toxicity. A specific paradox revealed in our study is the role of manganese (Mn). It exhibited the highest percentage of regulatory exceedances, ostensibly posing the primary water quality problem in Polish groundwater. However, the risk assessment proved that Mn is not the primary driver of the hazard, and at numerous points, despite exceeding standards, it does not generate a health risk. This discrepancy highlights the decoupling of regulatory compliance from actual toxicological risk. Although Mn frequently exceeds drinking water standards across the network, its high oral reference dose (RfD = 0.14 minimizes its individual HQ. In contrast, elements with much lower RfD values (e.g., As) or those showing extreme localized concentrations (e.g., nitrates) dominate the non-carcinogenic risk. Furthermore, Mn does not possess an oral cancer slope factor (SF), meaning it contributes nothing to the TCR metric. This reinforces that many statutory thresholds for Mn protect aesthetic or organoleptic properties rather than reflecting toxicological boundaries. Conversely, the presence of arsenic (As), with its low RfD value (0.0003 mg/kg·day), implies that even minimal exceedances of the standard lead to a sharp increase in risk. This situation creates a false alarm; well users focus their attention on the ubiquitous manganese, while substantially more hazardous trace elements remain overlooked. The analysis of spatio-temporal variability at the groundwater body (GWB) level revealed a distinct contrast between a stable, geogenic background and points of high anthropogenic health risk. Manganese contamination exhibits the characteristics of a chronic and stable phenomenon linked to the geological background. In contrast, nitrates (NO-3) and boron (B), originating from anthropogenic sources (municipal wastewater discharges, agriculture, runoff), account for extreme and episodic concentration spikes that drive the highest HI values. Notably, nitrates are traditionally monitored for specific risks to infants (methemoglobinemia), but they are rarely integrated into the general HI as an equivalent non-carcinogenic risk factor for other age groups. Our study demonstrates that extreme NO-3 and B concentrations constitute the dominant risk factor, particularly for the pediatric population. The application of a probabilistic approach (Monte Carlo) demonstrated a fundamental flaw in reporting averaged risk values. While the median values (representing a typical consumer) suggested a safe exposure level, the 95th percentile analysis revealed alarming risks for the most sensitive segment of the population, especially children. It is also critical to delineate local hotspots where the HI for children reaches critical values (e.g., GWB 168, GWB 125). These findings indicate that, within our conservative modeling assumptions and ingestion-only exposure framework for the selected contaminants, regulatory compliance does not inherently ensure chemical safety. Such extreme localized risks threaten rural communities dependent on unmonitored private wells. Similar anthropogenic impacts and geochemical anomalies distort baseline water quality indices (WQI), amplifying risks for adjacent habitats (Karadeniz et al., 2024; Tokatlı et al., 2026; Yazman et al., 2026). Incorporating targeted health risk assessments alongside probabilistic Monte Carlo modeling provides a realistic framework for water resource management, decoupling regional geogenic baseline conditions from severe localized anthropogenic pressures.

Our findings have significant implications for environmental and health policies. The current classification of the chemical status of GWBs (PIG-PIB, 2024), based on environmental criteria, does not account for the direct use of these waters for consumption purposes. Water possessing a good chemical status may still pose a serious threat to public health. The results clearly highlight this discrepancy. According to the Water Framework Directive, this status is determined by a test based on environmental threshold values. This methodology averages the results on a spatial scale, thereby masking local hazards. This regulatory disconnect becomes even more pronounced when we view groundwater not simply as an isolated endpoint, but as a critical intermediate environmental medium. In real-world scenarios, groundwater functions as a receiving medium for pollutants originating from upstream sources—such as agricultural soils, pesticide application, and surface water infiltration. The anthropogenic spikes driving the “silent risk” in our study are the downstream result of this broader contaminant transport chain. Yet, current regulations treat these compartments in isolation. Soil quality standards, surface freshwater standards, groundwater thresholds, and drinking water limits are rarely harmonized. Consequently, an agricultural site might meet soil standards, and a nearby river might pass its ecological assessment, while continuous leaching of sub-threshold pollutant loads still accumulates in the underlying groundwater. When this water is later extracted from a private well, it becomes the final human exposure medium. Addressing mixture toxicity therefore requires moving beyond isolated drinking water comparisons. It underscores the urgent need for a coherent source-to-exposure framework—one that aligns environmental standards across soils, surface waters, and groundwater to prevent cumulative health hazards at the point of consumption. Similarly, the median values (HI-Median) from the Monte Carlo simulation frequently suggest a safe exposure level. Only the 95th percentile analysis (HI_P95) reveals the actual threat to vulnerable groups. The official chemical status assessment in 2022 assigned GWB No. 21 a good status. Meanwhile, our simulation showed a drastic increase in risk for children, with the HI_P95 value rising from 5.05 in 2010 to 16.48 in 2022. The official report notes contamination across 34.15% of the area (particularly in the shallowest aquifers). Regulations consider this extent to be local. From a public health perspective, this represents a clear underestimation. This 34.15% of the area constitutes a zone of unacceptable toxicological risk, particularly for children drinking water from private wells in this territory.

A convergence between the official assessment of a poor chemical status and a high health risk occurs less frequently. An example is GWB No. 43. The model confirms a stable and high health risk throughout the entire period. According to the official report, groundwater quality exceedances occur across 61.18% of the area. However, the official test fails to account for the additive hazard resulting from the mixture of geogenic and anthropogenic contaminants. Thus, while the official assessment indicates a poor chemical status, it does not reflect the actual scale of the public health threat. In 2022, GWB No. 9 and GWB No. 34 received a good chemical status. In GWB 9, the HI_P95 value for children according to the Monte Carlo model rose to 3.30, and in GWB 34, it reached 4.43. These values clearly demonstrate the silent risk.

In conclusion, a good chemical status of a GWB under current law does not signify safe drinking water. Present regulations do not protect the most vulnerable users, exposing them to “silent risk” due to their reliance on private water supplies. The discrepancy we have demonstrated between legal compliance and actual health risk proves that Health Risk Assessment (HRA), incorporating mixture toxicity, should constitute an additional, standard component of monitoring programs, particularly in areas with complex chemical characteristics and a high proportion of private groundwater wells.

Assuming that total chromium consists entirely of Cr(VI) introduces uncertainty and represents a limitation of this study. The national monitoring dataset reports only total chromium without speciation. Therefore, we could not separate benign Cr(III) from carcinogenic Cr(VI) and treated total chromium as 100% Cr(VI). This conservative approach protects public health and aligns with regulatory principles. However, the calculated Total Cancer Risk (TCR) for chromium is likely overestimated. In reality, a significant portion of groundwater chromium occurs as less toxic Cr(III) (Zhitkovich, 2011; Rakhunde et al., 2012). Future macro-scale assessments should include chromium speciation to refine exposure models and eliminate this risk inflation.

Another limitation is the omission of the dermal exposure pathway (bathing, showering). Since certain metals can be absorbed through the skin, the actual total exposure, and thus the risk, may be higher than estimated in this study (Eid et al., 2024; Haider et al., 2025). Furthermore, due to the absence of specific data, 100% gastrointestinal bioavailability was assumed for all analyzed elements. This is a standard but simplifying assumption in US EPA models that can lead to an overestimation of the actual absorbed dose. Despite these limitations, the conservative approach applied ensures that the identified risks are not underestimated, which is preferred in the protection of public health.

Future perspectives

Quantitative models provide a clear snapshot of localized health risks, but protecting public health over the long term requires looking at the bigger picture. Groundwater sit right in the middle of environmental chain. It serves as a vital bridge connecting upstream human activities and surface pollution with downstream human exposure. Because of this interconnectedness, evaluating groundwater quality purely through the lens of standalone drinking water standards creates a narrow view that can leave communities vulnerable.

A classic example of this environmental continuum is the lifecycle of agricultural pesticides and nutrients. When fertilizers or crop protection chemicals are applied to fields, they do not stay in one place. They move through the agricultural soil matrix, enter surface runoff, and seep into local freshwater systems before eventually infiltrating the water table. Once these contaminants reach an aquifer, they can quietly compromise drinking water wells and other domestic supplies. Recent work by Tang et al. (2025) underscores the critical importance of tracing this complete pesticide flow chain, from the farm field through freshwater systems to the tap. Furthermore, as highlighted by Li (2018) and Huang and Li (2025a, 2025b), there is an urgent need to harmonize water quality standards across these different compartments. When soil guidelines, surface water regulations, and groundwater standards do not align, we create regulatory blind spots that mask cumulative ecological and human health risks. Moving toward a cohesive “source-to-exposure” framework is essential if we want to catch and manage these shifting pollution pathways early.

At the same time, we need to broaden how we think about how people actually use water in their daily lives. Our current risk assessment focused strictly on the ingestion pathway because it is the most direct and conservative metric for contaminants like nitrates and boron. However, families relying on private wells use this groundwater for much more than just for a drink. They use it to cook, wash dishes, bathe their children, and fill backyard swimming pools. These routine activities open up additional exposure pathways, such as skin absorption during a bath or inhaling aerosolized compounds during a hot shower. While ingestion remains the primary concern for inorganic mixtures, future public health models should incorporate these multi-route pathways into probabilistic Monte Carlo simulations. Expanding our focus to include both multi-media environmental transport and diverse domestic water uses will help environmental scientists and policymakers fully unmask the "silent risk," ensuring safer water and better health protection for vulnerable communities.

Conclusion

This study provides a comprehensive, spatio-temporal health risk assessment associated with groundwater contamination. While these models serve as crucial theoretical screening tools, they have highlighted the inadequacies of the current regulatory frameworks in protecting public health. The main conclusions can be summarized as follows:

1. The most significant finding of this work is the identification of the “silent risk” phenomenon. Relying exclusively on compliance with legal standards to evaluate groundwater safety is insufficient. Water that meets legal requirements for individual contaminants can still generate unacceptable non-carcinogenic health risks (HI > 1.0) due to the additive effect of accumulating multiple substances present below their threshold values. Current legislation ignores mixture toxicity. However, it is essential to emphasize that exceeding the HI > 1.0 threshold is a conservative screening indicator of potential exposure, not a direct validation of actual adverse health outcomes or clinical harm.

2. The study emphasizes that children modeled within this framework bear a non-carcinogenic risk burden that is more than twice as high as that of adults. Water deemed legally safe for adults (HI < 1.0) frequently exceeds the safety threshold for the pediatric population, revealing a serious gap in the protection of the most vulnerable groups at the level of exposure estimation.

3. The most frequently exceeded parameter is manganese (Mn, > 60% of points), which is a result of a stable geological background. However, the high frequency of manganese exceedances creates a false alarm, distracting attention from the actual risk drivers (HI > > 1.0), which are localized, extreme anthropogenic contaminants, particularly nitrates and boron.

4. Carcinogenic risk (TCR) exceeds safe thresholds even at monitoring points that comply with legal standards. In contrast to non-carcinogenic risk, this stems from the fact that the toxicological cancer slope factors (SF) for heavy metals (As, Cr, Pb) are more restrictive and sensitive to low concentrations than the threshold values defined in regulations. As a probabilistic metric, an elevated TCR indicates a higher statistical likelihood of risk, but it does not validate or confirm the actual incidence of cancer in the studied populations.

5. The spatio-temporal analysis demonstrated a discrepancy between the official good chemical status of GWBs (under the Water Framework Directive) and the actual health risk. The stable geogenic background (Mn) masks localized, episodic risk spikes driven by anthropogenic contaminants. GWB units classified as safe can harbor hotspots of cumulative hazards, proving that environmental indices are insufficient for evaluating drinking water safety.

6. The demonstrated discrepancy proves that Health Risk Assessment (HRA), accounting for the cumulative effect of contaminants and the sensitivity of specific age groups, should become a mandatory standard for evaluating water intended for human consumption. However, a balanced interpretation requires acknowledging that HRA is a proactive early-warning system rather than a measure of confirmed clinical morbidity. Without recognizing cumulative exposure to contaminant mixtures, current regulations cannot fully guarantee public health safety, leaving users of private wells-who are most likely to consume untested water-particularly vulnerable to the identified “silent risk”. Therefore, integrating HRA into routine monitoring provides a scientifically robust method to prioritize targeted interventions for these vulnerable groups, even though it does not serve as a direct epidemiological diagnosis of actual health outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J. K. performed all stages of the research work and manuscript preparation.

Funding

The author declare that no funds, grants, or other support were received during the preparation of this manuscript.

Data availability

The datasets analysed during the current study are publicly available https://mjwp.gios.gov.pl/wyniki-badan/wyniki-badan-2025.html.

Declarations

Competing interests

The author have no relevant financial or non-financial interests to disclose.