Heavy metal pollution and health risk assessment in surface waters impacted by pharmaceutical effluents in Gazipur and Narayanganj, Bangladesh

Abstract

Surface water contamination by heavy metals (HMs) is a growing concern in industrial regions of Bangladesh. This study assessed HM levels and associated health risks in surface water near pharmaceutical-industrial zones in Gazipur and Narayanganj. Twelve samples were collected from the Turag and Shitalakshya rivers and analyzed for ten metals (As, Pb, Cd, Cr, Ni, Cu, Zn, Hg, Fe and Mn) using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Pollution levels were evaluated using the Heavy Metal Pollution Index (HPI), Heavy Metal Evaluation Index (HEI), Water Pollution Index (WPI) and Heavy Metal Toxicity Load (HMTL). Non-carcinogenic and carcinogenic risks were estimated for adults and children through ingestion and dermal exposure. Mean concentrations of Pb (153 ppb), Fe (1380 ppb), and Hg (7 ppb) exceeded WHO guideline values at most sites, and several locations showed HPI and WPI values above 100, indicating critical pollution. Hazard quotients for Pb and Cd were greater than 1 in children, and cancer risk from As and Cr was above the acceptable range. Multivariate statistical analyses such as principal component analysis and hierarchical clustering suggested pharmaceutical effluents as a dominant source. Correlation analysis supported the presence of common contamination origins, notably a strong positive correlation between Fe and Hg (r = 0.89) and between Pb and Ni (r = 0.94). These findings highlight the need for stricter effluent control and regular monitoring of surface water in industrial areas.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-39794-9.

Introduction

Water is one of the most vital natural resources for life on Earth. In Bangladesh, a riverine country with more than 700 rivers, water sustains agriculture, supports livelihoods, and shapes its geography^1,2. Around 70% of the population depends on agriculture, with water playing a key role in cultivating high-demand crops such as rice and jute^3,4. Over 80% of people rely on groundwater for drinking and daily activities⁵. However, access to clean and safe water remains a major challenge, especially in developing countries like Bangladesh. According to the World Health Organization (WHO), nearly one-third of the global population lacks access to safely managed drinking water⁶.

In rural Bangladesh, waterborne diseases caused by poor sanitation and unsafe water are responsible for a significant portion of health-related deaths⁷. According to the WHO, unsafe drinking water, inadequate sanitation, and poor hygiene collectively account for over 1.4 million deaths annually and remain a leading contributor to disease burden in low- and middle-income countries⁸. The rapid pace of urbanization, industrial growth, and population increase in Bangladesh has intensified pressure on the country’s water resources. Dhaka, with over 24.7 million people living within a ~ 306 km² area, ranks among the most densely populated cities globally^9,10. Industrial activity in the city’s outskirts has also expanded rapidly in the recent years¹¹. The Department of Environment, Bangladesh, has identified more than 450 polluting industries across Dhaka Division, including Gazipur and Narayanganj, where textile, dyeing, and pharmaceutical industries are heavily concentrated^12,13.

The pharmaceutical sector, while essential for healthcare, contributes significantly to environmental pollution. These industries consume large volumes of water and discharge the untreated remainder as effluent, often containing heavy metals and other contaminants¹⁴. Such effluents- complex mixtures of biologically active agents, solvents, and heavy metals, are frequently released into surface water with minimal or no treatment¹⁵. Consequently, rivers such as Buriganga, Turag, Shitalakshya, and their tributaries receive approximately 60,000 m³ of hazardous industrial waste daily¹⁶.

Heavy metal contamination in surface water has emerged as a critical environmental and public health concern. In Bangladesh, heavy metals (HMs) such as Pb, Cd, Cr, As, Ni, and Zn are found in urban rivers, lakes, agricultural soils, and food chains¹⁷. These potentially toxic elements (PTEs) are known for their ecological and human health hazards due to their high density, persistence and bioaccumulative properties. For instance, studies have reported elevated levels of Pb in mangoes and Cd in tomatoes collected from various agro-ecological zones, exceeding WHO and FAO safety limits¹⁸. Conversely, elements such as Cu, Fe, and Zn are essential micronutrients; however, their excessive concentrations can disrupt aquatic ecosystems and human health. High levels of Fe can damage plumbing and alter water taste, while elevated Zn can interfere with copper uptake in aquatic life. Excess Cu exposure may lead to liver and kidney dysfunction in humans¹⁹. Heavy metal contamination further extends to groundwater, vegetables, cereals, and freshwater fish species, all of which contribute to human exposure^20–23.

Long-term exposure to toxic trace metals such as arsenic and lead is known to cause serious health disorders. These include DNA damage, oxidative stress, and developmental abnormalities, with vulnerable groups such as children and pregnant women facing the greatest risk^24,25. The health risk is further compounded by the long biological half-lives of metals like cadmium, chromium, and lead, which allow them to accumulate in human tissues over time²⁶. To assess these risks, researchers use various pollution indices and health risk models. These tools help quantify pollution severity and identify both carcinogenic and non-carcinogenic hazards. However, simply comparing metal concentrations to reference levels is often inadequate for understanding long-term health implications²⁷. Therefore, this study aims to evaluate the contamination in surface waters near pharmaceutical industrial zones in Gazipur and Narayanganj by analyzing heavy metal concentrations. It employs a suite of indices, including the single factor pollution index (Pi), Nemerow pollution index (NPI), degree of contamination (Cd), heavy metal evaluation index (HEI), heavy metal pollution index (HPI), heavy metal toxicity load (HMTL), and water pollution index (WPI) to comprehensively assess pollution levels. Finally, the research estimates the associated health risks for both children and adults, considering both ingestion and dermal exposure pathways.

Although several studies from Bangladesh have reported heavy metal contamination in rivers receiving industrial discharges, important gaps remain. Most earlier works have concentrated on textile, dyeing or tannery industries, with limited attention to the rapidly expanding pharmaceutical-industrial zones^28,29. In Bangladesh, a few studies have examined pharmaceutical wastewater and effluent treatment performance and reported that pharmaceutical discharges often fail to meet environmental standards, posing risks to aquatic ecosystems and nearby communities^30,31. However, these studies largely focus on effluent physicochemical parameters and treatment efficiency rather than heavy metal contamination and associated health risks in receiving surface waters. Furthermore, previous research has often analyzed sediments or groundwater, which reflect legacy pollution^32,33. There is a critical lack of data on surface water, which presents the most direct and immediate exposure risk to human populations. Methodologically, most studies have applied either a small set of pollution indices or health-risk models alone, without integrating multimetric indices with multivariate statistics to identify potential sources. To the best of our knowledge, no previous study in Bangladesh has simultaneously focused on surface waters directly influenced by pharmaceutical effluents in both Gazipur and Narayanganj, applied a comprehensive suite of pollution indices alongside principal component analysis and hierarchical cluster analysis, and jointly evaluated non-carcinogenic and carcinogenic risks for adults and children. By addressing these gaps, the present study provides an integrated, pharmaceutical effluent-focused assessment of heavy metal pollution and associated health risks in Bangladesh, generating evidence that is directly relevant for discharge regulation, environmental monitoring and protection of vulnerable populations. Accordingly, this study had three specific objectives: (i) to quantify ten priority heavy metals in surface waters influenced by pharmaceutical effluents in Gazipur and Narayanganj using ICP-MS; (ii) to assess pollution status using a suite of indices (Pi, NPI, Cd, HEI, HPI, HMTL and WPI) together with multivariate statistical tools (principal component analysis and hierarchical cluster analysis); and (iii) to evaluate non-carcinogenic and carcinogenic health risks for adults and children via ingestion and dermal exposure. We hypothesized that sites receiving pharmaceutical effluents would show higher heavy metal concentrations, elevated pollution indices and greater health risks, and that multivariate analyses would group metals in patterns consistent with pharmaceutical-industrial sources.

Materials and methods

Study area and sample collection

Surface-water samples were collected from twelve key locations across two major industrial zones- Gazipur (Turag River) and Narayanganj (Shitalakshya River) between late November 2024 and early January 2025. These rivers receive effluents from textile, dyeing, chemical, and pharmaceutical industries. The area experiences a tropical monsoon climate with significant seasonal variations in rainfall and temperature, which influence pollutant concentration and dispersion^34,35. Site selection was based on proximity to pharmaceutical manufacturing plants and visible discharge outlets, identified through field reconnaissance (Figs. 12). Each sampling site represented a distinct effluent exposure profile: Discharge Point-1 (near outlet), Point-2 (midstream, ~ 100 m), and Point-3 (downstream, ~ 250 m). A brief description of the geographic coordinates of each site is provided in Table 1. Two replicate samples were collected per point, yielding a total of 24 surface-water samples. Water samples (500 mL each) were collected using pre-cleaned polyethylene bottles that were rinsed with site water prior to sampling. The bottles were immediately sealed in airtight polyethylene bags, placed in pre-chilled insulated ice boxes containing frozen gel packs to maintain a temperature of approximately 2–3 °C during transport and subsequently preserved at 4–8 °C at the Bangladesh Council of Scientific and Industrial Research (BCSIR) laboratory until analysis.

Fig. 1 .

Surface water (SW) collection points along the Turag River in Gazipur, Bangladesh. Map annotations generated by ArcGIS 10.7 under a valid institutional license. Base satellite imagery from Google Earth Pro (version 7.3.6.10441; Image © 2026 Airbus © Google).

Fig. 2.

Surface water (SW) collection points along the Shitalakshya river in Narayanganj, Bangladesh. Map annotations generated by ArcGIS 10.7 under a valid institutional license. Base satellite imagery from Google Earth Pro (version 7.3.6.10441; Image © 2026 Airbus © Google).

Table 1.

Geographic coordinates of surface-water sampling sites (SW-1 to SW-12) along the Turag and Shitalakshya rivers near major pharmaceutical-industrial installations in central Bangladesh.

River	Area	Sample no.	Latitude	Longitude
Turag	Gazipur	SW-1	23°53’49.92"N	90°23’09.83"E
		SW-2	23°53’46.54"N	90°23’12.16"E
		SW-3	23°53’44.57"N	90°23’13.56"E
		SW-4	23°53’28.42"N	90°23’23.83"E
		SW-5	23°53’25.65"N	90°23’27.80"E
		SW-6	23°53’23.65"N	90°23’29.41"E
Shitalakshya	Narayanganj	SW-7	23°40’18.47"N	90°31’52.71"E
		SW-8	23°40’13.26"N	90°31’49.76"E
		SW-9	23°40’07.12"N	90°31’43.89"E
		SW-10	23°38’24.40"N	90°31’08.62"E
		SW-11	23°38’20.29"N	90°31’06.43"E
		SW-12	23°38’15.69"N	90°31’07.24"E

Sample digestion and analytical procedure

Sample bottles were soaked in 10% nitric acid for 24 h and rinsed thoroughly with deionized water. Each 40 mL water sample was transferred into acid-cleaned PTFE vessels and digested with 5 mL HNO₃ (65%) and 2.5 mL H₂O₂ (30%) using microwave digestion. The digest was cooled, diluted to 50 mL with ultrapure water, and analyzed using ICP-MS (PerkinElmer NexION 2000, USA) for ten heavy metals: As, Pb, Cr, Cu, Ni, Zn, Hg, Cd, Fe, and Mn.

Quality assurance and quality control (QA/QC)

Comprehensive quality assurance and control protocols were rigorously maintained throughout the sampling, digestion, and ICP-MS analysis to ensure the integrity of the analytical data. To minimize contamination, all glassware and PTFE digestion vessels were first soaked in 10% ultrapure nitric acid for 24 h and then thoroughly rinsed with deionized water. The ICP-MS instrument (PerkinElmer NexION 2000, USA) was optimized daily using the manufacturer’s NexION Setup Solution prior to calibration. Calibration curves, established with a Merck ICP Multi-element Standard Solution XIII across a range of 1–100 ppb, confirmed a linear instrument response (R² > 0.9995). A mixed internal-standard solution containing ⁴⁵Sc, ⁷³Ge, ⁸⁹Y, ¹¹⁵In, ¹⁸⁵Re, ¹⁹³Ir, and ²⁰⁹Bi was introduced to all standards and samples to correct for matrix effects and instrumental drift. Instrument precision was maintained within a relative standard deviation (RSD) of 5–8% for all analytes based on triplicate measurements. The accuracy of the analysis was verified by processing a Standard Reference Material (SRM NIST 1643f: Trace Elements in Water) with each sample batch. All reported concentrations were corrected for blank values and recovery rates where applicable. The measured recoveries ranged between 95.97% and 104.21%, confirming the reliability of the analytical process. Finally, the detection limits (LOD) and quantification limits (LOQ) for each analyte were determined using a signal-to-noise ratio approach, defined as 3 times and 10 times the background signal, respectively. LOD and LOQ were sufficiently low to capture background concentrations of trace metals. For instance, LOD/LOQ (µg/L) values were: As 0.0124/0.123, Pb 0.029/0.31, Cd 0.0052/0.053, Cr 0.063/0.62, Ni 0.063/0.61, Cu 0.051/0.49, Zn 0.11/0.98, Hg 0.0013/0.012, Fe 0.12/1.02, and Mn 0.072/0.73.

Multivariate statistical analysis

All data were processed in triplicate and expressed as mean ± SD. Statistical analyses were conducted in SPSS v26 to assess inter-metal relationships, pollution sources, and clustering. Multivariate statistical approaches, specifically Correlational analysis, Principal component analysis (PCA), hierarchical cluster analysis (HCA), were employed to elucidate the interrelationships among variables (heavy metals), identify distribution patterns, and determine potential contamination sources in the study area. Correlation analysis, based on Pearson’s correlation coefficients, assesses linear relationships, ranging from − 1 to 1. Positive correlations indicate direct associations, whereas negative correlations indicate inverse associations. The correlation matrix demonstrates significant associations, offering insights into correlated variables that may stem from shared sources, synergistic effects or similar environmental dynamics. For instance, strong correlations between variables suggest common anthropogenic sources (such as industrial effluents), whereas, weak correlations indicate independent or site-specific inputs.

PCA was used to minimize data dimensionality while preserving key information, converting original variables into independent principal components, prioritized by their variance contribution. These components reflect dominant factors such as industrial effluents or natural processes. PCA loadings reveal the relative influence of each variable (heavy metals) on these components, facilitating the recognition of source-specific patterns. Hierarchical cluster analysis classifies variables by their similarities using Ward’s method with Euclidean distance. The resulting clusters often correspond to shared pollution sources or similar environmental conditions. Metals classified in the same cluster reflect shared transport pathways or common sources. Collectively, these analytical approaches provide a comprehensive framework to assess interrelationships, factors influencing variability, and analyze spatial-temporal trends, especially in environmental studies such as surface water quality evaluation.

Single-factor pollution index (Pi)

Using the single-factor index approach, water integrity classification is based on the most critical pollutant, its measured concentration is evaluated relative to the WHO water quality standard, enabling the identification of key contaminants and their hazard levels in a specific water body. This method is particularly useful for pinpointing which single pollutant most significantly deviates from standard limits. To calculate the pollution index value, its observed concentration is divided by the allowable threshold defined by national or international water quality guidelines. The following Eq. (1) represents the single-factor pollution index.

1

Using the relationship above, Pi denotes the pollution index for a specific water quality parameter, Ci is the contaminant concentration (mg/L), and Si is the guideline value (mg/L), the pollution status is classified into five categories: Pi < 1 (Type I- unpolluted), 1–2 (Type II- slightly polluted), 2–3 (Type III- lightly polluted), 3–5 (Type IV- moderately polluted), and > 5 (Type V- heavily polluted)^36,37.

Nemerow pollution index (NPI)

A weighted multi-parameter index, the Nemerow method accounts for overall average pollution as well as the maximum contaminant level detected. This approach provides a more comprehensive assessment of water quality by considering both typical and extreme pollutant concentrations³⁸. Equation (2) expresses the Nemerow pollution index:

2

In the formula, the composite pollution level (PN) at a sampling point is calculated from the maximum (Pimax) and average (Pi) single-factor indices. Interpreting PN yields these classifications: below 1 indicates standard water quality; 1–5 indicates slight contamination; 5–10 indicates moderate pollution; and values above 10 signal severe pollution³⁹.

Degree of contamination (Cd)

This method applies the contamination factor, determined by the ratio of each heavy metal concentration to its background level, and the degree of contamination, which sums all C_f values, to evaluate HM pollution in environmental water samples. This approach compares the concentration of each HM to a predefined standard or background value, offering a precise understanding of pollution severity³⁹. This is expressed through the following equations:

3

4

The contamination factor for the i^th heavy metal is given above (Eq. 4). Summing these yields the degree of contamination (C_d). Based on Hakanson’s criteria, a C_d value less than 1 indicates minimum contamination, C_d factor between 1 and 3 indicates moderate contamination, and C_d value greater than 3 signifies high contamination³⁶.

Heavy metal evaluation index (HEI)

HEI is a straightforward metric that gauges overall water quality by adding up each heavy metal’s concentration relative to its regulatory threshold, reflecting their combined effects on human health and suitability for drinking water^36,40,41. To derive the HEI, each heavy metal’s concentration in the sample is expressed as a fraction of its corresponding maximum permissible value, and these values are summed, as shown in Eq. (5).

5

In the HEI equation, n denotes the total metals analyzed; Ci and MACi are the measured concentration and regulatory limit for each metal, respectively. Based on HEI values, water quality is divided into low (HEI < 10), medium (HEI 10–20), and high (HEI > 20) contamination categories⁴².

Heavy metal pollution index (HPI)

In this approach, overall water quality is assessed based on the presence of contaminated trace metals using the Heavy Metal Pollution Index (HPI). The process calculates HPI by combining weighted sub-indices for each metal (with weights between 0 and 1) using a formula involving both Q_i and W_i, as outlined in Eqs. (6) and (7)^43,44.

6

7

The variable n represents the total trace elements evaluated; Wi is the unit weight for i^th metal, typically calculated inverse to its maximum acceptable concentration, and is used to weight the contribution of each pollutant in the overall HPI (such as As = 0.0372, Pb = 0.0372, Cd = 0.0744, Cr = 0.0074, Ni = 0.0186, Cu = 0.0002, Zn = 0.0007, Hg = 0.3721, Fe = 0.0019, Mn = 0.0074), Si is the maximum permissible limit of the i^th heavy metal, Qi is the sub-index of the i^th metal, and Ii is the maximum desirable limit of the i^th metal. The resulting HPI then falls within five quality categories: excellent (< 25), good (26–50), poor (51–75), very poor (76–100), and unsafe for drinking (> 100)⁴⁵.

Heavy metal toxicity load (HMTL)

This metric evaluates health-related metallic pollution levels in surface water and the treatment effort required to render it safe. Using Eq. 8, the concentration of each metal is multiplied by its ATSDR-assigned hazard score (HIS), with all results summed to yield the HMTL.

8

Where Mi is the observed concentration of a heavy metal (mg/L), n is the total count of trace metals, and HISi is the hazard intensity score of the trace metals. Such scores are: As = 1675, Pb = 1531, Cd = 1317, Cr = 892, Ni = 994, Cu = 807, Zn = 916, Hg = 1455, Fe = 0, Mn = 799⁴⁶. The Hazard Intensity Score integrates three components: exposure potential, chemical toxicity, and frequency of occurrence on the National Priorities List (NPL), to gauge the relative health risk posed by a pollutant.

Water pollution index (WPI)

Introduced by Horton in 1965, the WPI serves as a tool for assessing, monitoring, and regulating water pollution, with its value computed using Eq. (9)⁴⁷.

9

The variables Mi, Mini, and Ri define the metal’s average concentration, minimum limit, and acceptable range respectively. Water quality categories according to WPI are: < 0.5 for excellent, 0.5–0.75 good, 0.75-1.0 fairly polluted, and > 1.0 indicating severe pollution and unsuitability for human consumption⁴⁸.

Health risk assessment

To estimate the health implications of groundwater metallic contamination, the study applied Health Risk Assessment (HRA) methods, including calculating hazard quotient (HQ), lifetime average daily dose (LADD), and relevant empirical indices for both ingestion and dermal exposure pathways, assessing impacts on adults and children⁴⁹.

Non-Carcinogenic health risk

The non-carcinogenic risk assessment was conducted according to the methodology recommended by the United States Environmental Protection Agency²⁰. Non-carcinogenic effects were evaluated using the hazard quotient (HQ), which quantifies the estimated exposure relative to a reference dose (RfD). Exposure can proceed via two primary routes: oral intake and skin contact. The LADD through ingestion is calculated as:

10

Where: C = Concentration of heavy metal (mg/L), IR = Ingestion rate (L/day), EF = Exposure Frequency (365 days/year), ED = Exposure Duration (6years for children, 30 years for adults), BW = Body Weight (16 kg for children, 70 kg for adults), AT = Averaging Time (ED × 365).

To ensure comparability with previous United States Environmental Protection Agency (USEPA) based assessments, all exposure parameters (ingestion rate, exposure duration, body weight, exposure frequency and averaging time) were adopted from standard guideline values for adults and children, and hazard quotient/carcinogenic risks were calculated as point estimates without additional probabilistic or sensitivity analysis of these assumptions.

The LADD through dermal contact is calculated as:

11

where: ET = Exposure Time (1 h/day for Children, 0.58 h/day for Adults) Kp = Skin permeability coefficient (Cd = 0.001, Cr = 0.002, Cu = 0.001, Fe = 0.001, Mn = 0.001, Ni = 0.0002, Pb = 0.0001, Zn = 0.0006 cm/h), SA = Skin surface area exposed (600 cm² for children, 1800 cm² for adults), CF = Unit conversion factor (0.001 L/cm³) Other variables are as previously defined.

The Non-carcinogenic Hazard Quotient (HQ) is then determined using:

12

Where RfD is the reference dose for the respective metal (mg/kg-day).

13

Carcinogenic health risk

For carcinogenic risks (CR), the lifetime probability of cancer development resulting from exposure to heavy metals can be estimated by Eq. (14).

14

Where CSF is the cancer slope factor (mg/kg/day), specific to each heavy metal. CR values in the range of 10^-6 to 10^-4 are generally considered acceptable or tolerable⁵⁰. The calculations were performed separately for oral intake and dermal pathways in Children and Adults, using corresponding physiological and exposure parameters. Reference values for RfD and CSF are provided in Supplementary Table 1.

Results

Quantitative analysis

Findings indicate significant heavy metal pollution in the Turag and Shitalakshya rivers, both of which are heavily exposed to effluents from industrial zones in Gazipur and Narayanganj. Presented in Table 2 are the concentrations of heavy metals such as arsenic, lead, cadmium, chromium, nickel, copper, zinc, mercury, iron, and manganese in the water samples from different points. The measured values, expressed in parts per billion (ppb), are compared to the permissible limits set by the WHO⁵¹.

Table 2.

Observed concentration ranges of heavy metals compared to standard permissible limits.

Analyzed heavy metals	Range of concentrations of measured HMs (in ppb)	Standard permissible limit (in ppb)52	The number of samples observed exceeding the WHO standard
As	2.46–10.45	10	1
Pb	5.68-449.57	10	10
Cd	0.05–2.56	3	0
Cr	4.19–41.13	50	0
Ni	6.71–32.2	70	0
Cu	8.09–140.80	2000	0
Zn	16.07-505.88	3000	0
Hg	0.53–33.70	6	6
Fe	214.4-2284.05	300	10
Mn	103.05-817.84	400	4

Correlation analysis

The correlation matrix (Fig. 3) displays both the numerical and graphical representations of Pearson correlation coefficients among ten heavy metals (As, Pb, Cd, Cr, Ni, Cu, Zn, Hg, Fe, and Mn) in surface water samples. The strength and direction of these correlations were visually represented using colored circular symbols, where deeper shades of blue denote strong positive correlations, and red indicates negative relationships. Several strong and statistically significant positive correlations were observed, notably between Pb and Zn (r = 0.94), Cr and Pb (r = 0.92), Cr and Zn (r = 0.91), and Cr and Ni (r = 0.89). These associations suggest shared anthropogenic sources such as electroplating, industrial discharge, or vehicular emissions. Cd also exhibited strong correlations with Zn (r = 0.802) and Fe (r = 0.798), indicating a possible co-occurrence through smelting, waste disposal, or runoff from industrial zones. Fe was strongly correlated with multiple metals, including Zn (r = 0.95), Cr (r = 0.89), and Pb (r = 0.82), implying complex interactions, and possible mixed origins from both industrial activities and natural mineral dissolution. Cu showed a strong correlation only with Hg (r = 0.89), suggesting a distinct source profile. In contrast, Mn and As showed generally weak correlations with other metals (e.g., As–Mn r = 0.08), indicating separate geochemical behaviors, or lesser influence from shared pollution sources. These weak associations imply natural sources or isolated anthropogenic inputs.

Fig. 3.

Pearson correlation matrix (r) of HM concentrations (As, Pb, Cd, Cr, Ni, Cu, Zn, Hg, Fe, Mn) in surface water. The color scale represents the Pearson correlation coefficient (dimensionless), ranging from − 1 (strong negative correlation) to + 1 (strong positive correlation).

Principal component analysis (PCA)

Principal Component Analysis (PCA) was applied to identify latent patterns and potential pollution sources of HMs in the surface water samples collected from the study area. This multivariate technique reduces the dimensionality of complex datasets while preserving the variance structure, thereby enabling a clearer interpretation of inter-metal relationships and their environmental implications. In this context, PCA assists in distinguishing different contributing sources of HM contamination, such as industrial discharge, vehicular emissions, and geogenic inputs. As shown in Table 3, three principal components (PCs) with eigenvalues greater than 1 were extracted following the Kaiser criterion. These three components collectively accounted for 88.553% of the total variance in the HM dataset, signifying a substantial representation of the underlying structure. The scree plot (Supplementary Fig. 1) revealed a sharp decline in eigenvalues after the third component, reinforcing the suitability of retaining three components for meaningful interpretation.

Table 3.

PCA with total variance explained for all HMs found in the studied area.

Component	Initial Eigenvalues			Extraction sums of squared loadings			Rotation sums of squared loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	5.639	56.392	56.392	5.639	56.392	56.392	5.110	51.099	51.099
2	2.249	22.488	78.880	2.249	22.488	78.880	2.446	24.462	75.561
3	0.967	9.673	88.553	0.967	9.673	88.553	1.299	12.992	88.553
4	0.580	5.802	94.355
5	0.350	3.503	97.859
6	0.119	1.190	99.048
7	0.075	0.746	99.794
8	0.013	0.129	99.922
9	0.007	0.073	99.996
10	0.000	0.004	100.000

The first principal component (PC1) had an initial eigenvalue of 5.639 and explained 56.392% of the total variance. After Varimax rotation, PC1 accounted for 51.099% of the variance, making it the dominant factor influencing HM distribution. This component exhibited strong positive loadings for Zn (0.983), Cr (0.955), Pb (0.950), Fe (0.938), Ni (0.880), and Cd (0.791). The high intercorrelations among these metals suggest a shared origin, most likely from anthropogenic activities such as industrial emissions, leather tanning, battery waste, smelting, and effluents from metal processing plants. Chromium and lead are commonly linked with traffic-related emissions and industrial activities, while nickel and cadmium may be associated with electroplating, alloy production, and phosphate fertilizers. The substantial contributions of iron and zinc may stem from both anthropogenic and natural sources, including vehicular wear, industrial coatings, and sediment resuspension.

The second principal component (PC2) had an eigenvalue of 2.249 and explained 22.488% of the variance before rotation, increasing to 24.462% after rotation. This component showed high loadings for Cu (0.962) and Hg (0.922), and to a lesser extent, As (0.724). These metals may originate from specific industrial processes such as metal plating, mining runoff, chemical industries, and fossil fuel combustion. Copper and mercury are often linked to the production of pesticides, electronics manufacturing, and improper disposal of industrial waste. The presence of arsenic may suggest contamination from both natural mineral leaching and anthropogenic sources like tanneries or pesticide applications.

The third principal component (PC3) accounted for 9.673% of the variance prior to rotation and 12.992% after rotation. While the component accounted for a smaller portion of the overall variance compared to PC1 and PC2, it still reflects an important dimension of variation in the HM dataset. In the rotated matrix, PMn (0.288) showed moderate loading, suggesting influence from natural geochemical processes and weak anthropogenic activity. This distinct loading of Mn on PC3 likely reflects its strong dependence on redox conditions at the sediment-water interface rather than direct inputs from pharmaceutical effluents. Mn is primarily derived from the weathering of silicate and carbonate minerals, and its concentration in surface water often increases through reductive dissolution of Mn oxides under low-oxygen or slightly acidic conditions. At the sediment-water boundary, organic matter decomposition can lower redox potential, leading to microbial or chemical reduction of Mn (III/IV) oxides to the more soluble Mn (II) form, which can diffuse into the overlying water; subsequent re-oxidation under more oxic conditions promotes re-precipitation and re-adsorption to particles. Such redox-driven Mn cycling at the sediment-water boundary has been widely documented in lakes, rivers and estuaries and can generate spatially patchy but persistent background enrichment, partly decoupled from the anthropogenic metal patterns captured in PC1 and PC2^53–55. In the study area, shallow sediment layers along the Turag and Shitalakshya rivers contain Mn-bearing minerals such as pyrolusite and rhodochrosite, which may contribute to baseline Mn levels even in sites less impacted by direct effluent discharge. Similar geogenic enrichment of Mn was reported in the Shitalakshya River sediments by Kabir et al. (2020)³⁴ and in the Yellow River basin by Li et al.³⁹.

The rotated component matrix (Supplementary Table 2) provides a simplified structure that facilitates clearer interpretation by minimizing complex overlaps between variables. This statistical simplification reveals meaningful groupings that can be linked to distinct pollution sources. For instance, PC1 may represent a composite influence of industrial and traffic-related emissions, PC2 may reflect heavy metals from specialized industrial discharge and chemical inputs, while PC3 suggests contributions from natural sources or less intensive anthropogenic activities.

The 3D component plot in rotated space (Supplementary Fig. 2) visually illustrates the clustering of metals based on their principal component scores. Metals located close together in this plot are likely influenced by similar sources. In addition, the dendrogram generated through hierarchical cluster analysis (Supplementary Fig. 2) supports these findings by grouping metals with analogous loading patterns and environmental behaviors. The similarity between PCA-based groupings and hierarchical clustering strengthens the reliability of the source apportionment results.

Cluster analysis

Hierarchical Cluster Analysis (HCA) was performed to explore the grouping behavior of the ten analyzed HMs across the twelve surface water sampling sites. Using Ward’s linkage method and Squared Euclidean Distance as the similarity measure, a dendrogram was generated (Supplementary Fig. 2) to illustrate the degree of association among the metals.

The dendrogram revealed three distinct clusters at a rescaled distance of approximately 10. The first major cluster comprised Cu, Zn, Cr, Ni, Cd, As, and Mn, indicating a strong interrelationship and potentially common anthropogenic sources. These metals are frequently associated with industrial effluents, particularly from pharmaceutical and chemical manufacturing units, and tend to exhibit similar behavior in aquatic environments. The second cluster included Pb and Fe, which showed a moderate level of similarity. This grouping may suggest a partially shared source, possibly from corroded pipelines, industrial discharge, or environmental leaching, but with distinct concentration profiles compared to the first cluster. The third and most distinct cluster was formed by Hg, which displayed a higher rescaled distance from the other elements, indicating a unique spatial distribution pattern. This separation could be attributed to its specific industrial origin or different environmental transport and bioaccumulation characteristics.

Single factor pollution index (Pi)

By applying Eq. (1) to derive the Single Factor Pollution Index (Pi), the average levels in the samples were determined: Arsenic (As) 0.58, Lead (Pb) 5.29, Cadmium (Cd) 0.24, Chromium (Cr) 0.23, Nickel (Ni) 0.20, Copper (Cu) 0.02, Zinc (Zn) 0.03, Mercury (Hg) 1.59, Iron (Fe) 2.55, and Manganese (Mn) 0.83. Water quality ratings classify mercury as slightly polluted, iron as lightly polluted, and lead as heavily polluted. Pi values for the remaining metals fall below 1, signifying no notable pollution from arsenic, cadmium, chromium, nickel, copper, and zinc.

Nemerow pollution index (NPI)

The computed NPI values of the surface water using Eq. (2) span from 0.93 to 32.06, with a mean of 4.55 (shown in Supplementary Table 3), which, according to the NPI classification, indicates that the surface water is moderately polluted with different HMs.

Degree of contamination (Cd)

The Degree of Contamination for this research region ranges from − 6.19 to 49 with a mean of 1.57 (Supplementary Table 3), which indicates a medium degree of contamination in this surface water. The numbers of samples are classified according to Edet et al.³⁶, where 9 of them show a low degree of pollution, except for SW1, SW4, and SW10 which show moderate pollution and a high degree of pollution, respectively (Table 4).

Table 4.

Assessing water quality in the study area using a pollution‑index classification.

Pollution assessment index	Classification sources	Categories	Quality of the water samples	Number of samples	Samples observed in each class (%)
Cd	Edet et al.36	< 1	Low	9	75
		1–3	Moderate	1	8.33
		> 3	High	2	16.67
HEI		< 10	Low	8	66.67
		10–20	Medium	3	25
		> 20	High	1	8.33
HPI		< 25	Excellent	1	8.33
		26–50	Good quality	4	33.33
		51–75	Poor	0	0
		76–100	Very poor	2	16.67
		> 100	Unconsumable	5	41.67

Heavy metal evaluation index (HEI)

Using Eq. (5), we calculated the Heavy Metal Index (HEI) for surface water of this study region. The calculated HEI for this region averaged 11.56, exhibiting a minimum of 3.23 and a maximum of 59 (Supplementary Table 3). According to Edet et al.³⁶ method, around 66.67% of sampling points fell under the low pollution category, 25% were considered moderately polluted, and 8.33% were rated as high pollution (Table 4).

Heavy metal pollution index (HPI)

The Heavy Metal Pollution Index (HPI) derived via Eqs. 6 and 7-averaged 151.50, spanning from a minimum of 14.81 to a maximum of 470.5 across sampled sites, as detailed in Supplementary Table 3. Among these, 5 samples exceeded the threshold of 100, indicating the water is not suitable for consumption. Based on HPI categories, 8.33% fall within < 25, 33.33% of samples fall within 26–50, showing good quality, while 0% indicate low pollution, 16.67% medium pollution, and 41.67% fall under high pollution (Table 4), suggesting that parts of the area are heavily polluted and unfit for use.

Heavy metal toxicity load (HMTL)

To estimate the toxic heavy-metal load across the study region, HMTL values were derived for each sample in accordance with Eq. 8. The Heavy Metal Toxicity Load (HMTL) of samples collected from the studied region ranges between 173.6 and 1651.5, with a mean value of 522.36 (Supplementary Table 3). Four samples, SW1, SW4, SW10, and SW11 surpassed the risk threshold of 500, indicating elevated toxicity. The results could point to possible hazards from ingestion for humans and adverse effects on the local ecosystem. The HMTL values across the sampling sites are ranked as follows: SW-4 > SW-1 > SW-10 > SW-11 > SW-7 > SW-2 > SW-5 > SW-8 > SW-9 > SW-3 > SW-12 > SW-6.

Water pollution index (WPI)

Water Pollution Index (WPI) of the study area water was assessed using Eq. (9). The results shown in the Supplementary Table 6 indicate that most of the elements have WPI values below 0.5, reflecting insignificant pollution and safe concentration levels except for a few heavy metals like manganese (Mn) that fall under the moderately polluted category, indicating elevated levels that may require monitoring. In contrast, lead (Pb), mercury (Hg), and iron (Fe) are classified as highly polluted, suggesting serious contamination that could pose risks to both ecological and human health.

Non-carcinogenic health risk

In evaluating non-carcinogenic risks among adults and children, the study considered heavy metals such as arsenic (As), lead (Pb), cadmium (Cd), chromium (Cr), nickel (Ni), copper (Cu), zinc (Zn), mercury (Hg), iron (Fe), and manganese (Mn). The data from various sampling stations in the study area, as detailed in Supplementary Tables 4 and 5, highlight the hazardous conditions posed by these individual heavy metals. If the value of Hazard Quotient (HQ) is below 1, then it suggests that the exposure level is within safe limits, whereas an HQ above 1 signals a potential health risk⁵⁵.

In order to estimate health risks via oral, the ingestion hazard quotient (HQ) was computed for both age groups by assessing the concentrations of heavy metals in the study area using Eqs. (10,11,12,13). The ingestion hazard quotients (HQ) for both adults and children are summarized inSupplementary Table 4. In adults, the total HQ values spanned from 0.35 to 6.06, with an average value of 1.47, while in children, the range is notably higher 1.55 to 26.52, with a mean value of 6.45 (Fig. 4a). These outcomes demonstrate that children are at significantly greater risk from ingestion-related exposure to heavy metals. To determine the possible health hazards associated with dermal exposure to heavy metal-contaminated water, the dermal Hazard Quotient (HQ) was calculated for both adults and children using Eqs. (10,11,12,13). As presented in Supplementary Table 5, the total dermal HQ values in adults ranging from 0.00 to 0.01, with a mean of 0.0026, while for children, the values range from 0.0014 to 0.0122, with a mean of 0.0072 (Fig. 4b). Similar to ingestion exposure, the results indicate that children are more vulnerable to dermal absorption of metals such as As, Pb, Cd, Cr, Ni, Cu, Zn, Hg, Fe, and Mn, compared to adults.

Fig. 4.

Box-and-whisker plots showing variation in non-carcinogenic hazard quotient (HQ, dimensionless) and carcinogenic risk (CR, dimensionless) for surface-water exposure across all sampling sites a HQ^Ingestion, b HQ^Dermal, c CR^Ingestion (Adult), d CR^Ingestion (Children), e CR^Dermal (Adult), and f CR^Dermal (Children).

Carcinogenic health risk

Carcinogenic risk was calculated for the intake of heavy metals via contaminated water, based on exposure dose and cancer slope factors to assess potential cancer-related health effects. Carcinogenic risk assessments express the likelihood of cancer from long-term exposure to a contaminant or several contaminants. Carcinogenic risks from ingestion exposure were assessed for both adults and children in water samples, focusing on heavy metals including As, Pb, Cd, and Cr. For adults, CR ingestion values were significantly lower: As: 5.27 × 10^-5 to 0.00022, Pb: 6.9 × 10^-7 to 5.46 × 10^-5, Cd: 4.36 × 10^-6 to 0.00022, Cr: 2.99 × 10^-5 to 0.000294 (Fig. 4c). For children, the ingestion CR for the heavy metals ranged from, As-0.00023 to 0.00098, Pb-3.02 × 10^-6 to 0.00024, Cd-1.91 × 10^-5 to 0.00098, and Cr-0.00013 to 0.00129 (Fig. 4d).

To evaluate cancer risks through skin contact with contaminated water, dermal carcinogenic risks (CR dermal) were determined for both adults and children using Eq. (14). The analysis focused on Arsenic (As) and Chromium (Cr), as all other assessed metals (Pb, Cd, Ni, Cu, Zn, Hg, Fe, Mn) showed zero carcinogenic potential through dermal exposure. Among children, CR dermal values for arsenic ranged from 3.19 × 10^-8 to 4.15 × 10^-7, while Cr ranged from 7.87 × 10^-8 to 7.71 × 10^-7 (Fig. 4e). For adults, arsenic values ranged from 1.27 × 10^-8 to 1.65 × 10^-7, and Cr from 3.13 × 10^-8 to 3.07 × 10^-7 (Fig. 4f).

Discussion

The present study highlights significant contamination of surface water by toxic heavy metals (HMs), with notably high concentrations of Pb, Fe, and Hg, exceeding WHO limits across most of the twelve sampling locations. This indicates severe degradation of water quality in the Gazipur and Narayanganj industrial zones. The findings are consistent with earlier studies on urban rivers of Bangladesh, where heavy metal pollution was linked to unregulated industrial discharge. For instance, Sarkar et al.²⁸ reported elevated levels of Pb, Cd, and Cr in the Turag River near Dhaka, largely from textile and dyeing industries. In contrast, the present study identifies pharmaceutical effluents as a major pollution source, as evidenced by the predominance of Pb, Fe, and Hg, elements commonly associated with pharmaceutical and chemical manufacturing residues. In the study area, these patterns are likely driven by a combination of poorly treated pharmaceutical and mixed industrial effluents, dry-season low flows, and limited dilution capacity of the rivers. Pharmaceutical wastewater often contains a mixture of active ingredients, process chemicals and trace metals such as Cr, Ni, Cu, Cd and As, together with corrosion products from pipelines and reactors, which can elevate Pb, Fe and other metals in receiving waters when treatment is inadequate^56,57. Reduced discharge during the dry season further concentrates these contaminants, explaining the consistently high Pb, Fe and Hg levels observed at sites closest to major discharge points^58,59. Both studies employed multimetric pollution indices (e.g., HPI) and health risk assessments, identifying Pb and Cd as the most concerning contaminants for children. However, the current study strengthens source attribution by integrating multivariate analysis (PCA and HCA), which directly links heavy metal clustering to pharmaceutical effluents.

The concentrations of Pb, Fe, Cr, and Cd observed here are consistent with findings from Hossain et al.⁶⁰, who documented comparable contamination in the Turag and Buriganga rivers, as well as with Rahman et al.⁶¹, who examined the Shitalakshya River in Narayanganj. While Rahman et al. reported higher Pb, Cd, and Ni levels, Fe concentrations were lower than those observed in this study. Both studies underscored industrial effluents as dominant pollution sources, corroborating the present results. Recent work in the Basan Industrial Area of Gazipur has similarly reported severe exceedances of WHO limits for Cd, Pb, Ni and Cr in surface waters receiving mixed industrial discharges, with health risk indices indicating that children are the most vulnerable group⁴⁹. Likewise, an integrated assessment of water, soil and vegetables from Narayanganj industrial zones showed that heavy metal contamination can propagate across environmental compartments and magnify ecological and human health risks for nearby communities⁶². The agreement of our data with those of Sharmin et al.⁶³, who reported Pb and As levels above WHO thresholds in Dhaka groundwater, further confirms the widespread presence of these metals across environmental compartments, emphasizing potential cross-contamination between surface and subsurface waters.

The contamination pattern identified in this study also aligns with the dry-season enrichment trends reported by Islam and Azam⁶⁴, who found elevated Fe, Pb, Hg, and Cr in three major rivers, including the Turag and Shitalakshya. This seasonal concurrence suggests reduced dilution and increased pollutant concentration during the dry months, a characteristic also observed in the present dataset. Comparable heavy metal enrichment was documented in the Sutlej River (India) by Setia et al.⁶⁵, where Fe, Cr, Cd, Pb, and Ni levels were elevated due to industrial effluents. While both studies identified industrial discharge as the key contributor, the current study uniquely implicates pharmaceutical effluents as a dominant source- contrasting with textile and tannery-driven contamination observed in India. Both investigations, however, reported non-carcinogenic and carcinogenic risk levels higher for children, underscoring consistent exposure vulnerabilities in industrial zones across South Asia. Comparable multi-index and health-risk approaches applied to rivers in China and other regions have also identified mixed industrial sources as dominant drivers of As, Sb, Pb and Hg risks, reinforcing that the patterns observed in Gazipur and Narayanganj are part of a broader global concern about industrially driven metal contamination⁶⁶.

Comparable correlation trends observed in this study further validates the shared anthropogenic origin of metals. Strong positive associations such as Pb–Zn (r = 0.94), Cr–Ni (r = 0.89), and Fe–Pb (r = 0.82) indicates common input from mixed industrial discharge and surface runoff. Similar correlation structures were reported by Hossain et al.⁶⁰ for the Turag and Buriganga rivers and by Rahman et al.⁶¹ for the Meghna estuary, both linking Pb-Zn and Cr-Ni associations to industrial effluents. In India, Setia et al.⁶⁵ found comparable Pb-Zn-Cr clustering in the Sutlej River, while studies from the Chinese Loess Plateau and Pearl River Basin also reported co-occurrence of Pb, Zn, and Cr^69,70. The weak As-Mn correlation observed here aligns with Kabir et al.³⁴, suggesting distinct geogenic origins or localized sources. These parallels reinforce that the Gazipur-Narayanganj contamination profile follows a global pattern of multi-metal, co-mobilization from anthropogenic emissions and industrial wastewater. Recent multivariate analyses in other river basins have also shown that, the first few principal components and major clusters are typically dominated by anthropogenic metals such as Pb, Cr, Ni and As, while secondary components reflect geogenic backgrounds or diffuse sources, which is consistent with the PCA and HCA structures obtained in this study⁶⁹.

Although minor variations were observed between PCA and HCA groupings, particularly in the case of Hg, these differences are methodologically consistent and environmentally interpretable. In PCA, Hg exhibited a strong loading under PC2, indicating its statistical association with Cu and As and suggesting shared input from chemical and pharmaceutical industrial discharge. However, in HCA, Hg formed a separate cluster, reflecting its distinct spatial distribution and behavior in the aquatic environment. Such divergence arises because PCA emphasizes covariance and shared variance among variables, whereas HCA groups metals based on Euclidean distance and similarity patterns in concentration. Given Hg’s unique geochemical properties such as high volatility, strong affinity for organic matter, and tendency for phase partitioning, it often displays localized enrichment independent of other metals. Similar PCA-HCA divergence in Hg grouping has been reported in previous riverine studies^34,67. Therefore, the isolated clustering of Hg in HCA does not contradict the PCA interpretation but rather underscores its distinct transport and accumulation behavior within the study area.

These results confirm that untreated pharmaceutical and mixed industrial effluents are the principal sources of heavy metal pollution in the studied regions, with minor contributions from geogenic processes. The integrated use of multiple pollution indices (HPI, HEI, WPI) and multivariate analyses (PCA, HCA) provides a more comprehensive understanding of contamination dynamics than previous single-metric studies. By demonstrating elevated exposure risks in children, it emphasizes the urgent need for stricter regulation of pharmaceutical wastewater discharge and sustainable water management practices in Bangladesh.

The CR estimates revealed that As, Pb, Cd and Cr are the principal contributors to potential cancer development via oral ingestion. For adults, CR_ingestion values were highest for Cr and Cd, followed by As and Pb, whereas for children CR_ingestion values were higher for As and Pb, followed by Cd and Cr. When compared with the USEPA’s acceptable lifetime cancer risk range (10⁻⁶–10⁻⁴)⁵¹, all CR_ingestion values for these metals exceeded the upper bound, particularly in children, indicating an elevated probability of cancer over a lifetime of exposure. CR_dermal was calculated only for As and Cr, because the carcinogenic potency of the other metals via dermal contact is considered negligible. All CR_dermal values for both adults and children fell within or below the USEPA acceptable risk range, suggesting that ingestion is the dominant carcinogenic exposure route in the study area. However, when considering cumulative risk across metals and pathways, especially for children, the combined hazard warrants critical attention and underscores the need for effective mitigation of heavy metal contamination in surface waters used by nearby communities.

In the context of the Turag and Shitalakshya rivers, these water-based HQ and CR values likely underestimate the total exposure experienced by nearby communities. Households living along these industrial corridors frequently consume fish and other aquatic products sourced from local rivers, and several studies from Bangladeshi river systems have documented substantial accumulation of Pb, Cd, Cr and As in fish tissues, often yielding dietary HQ and CR values above recommended thresholds for regular consumers^70–72. Consequently, the elevated child HQ_ingestion and CR_ingestion estimates derived solely from surface-water exposure in this study should be interpreted as conservative, minimum risk levels. Actual health risks may be higher when combined exposures from drinking, bathing, and consumption of contaminated fish and irrigated produce are taken into account, particularly for children and other vulnerable groups.

Beyond direct risks to human health, the observed heavy metal concentrations also have important ecological implications for the Turag and Shitalakshya river ecosystems. Heavy metals such as Pb, Cd, Cr, As and Hg are persistent, can bioaccumulate in aquatic organisms, and may induce oxidative stress, growth retardation, reproductive impairment and behavioral changes in fish and invertebrates⁷³. In Bangladesh and other tropical river systems, elevated metal levels in water and sediments have been linked to accumulation in fish tissues and potential disruption of riverine food webs^70,71,74. Our findings of consistently high Pb, Fe and Hg, together with co-occurrence of other metals, suggest that chronic exposure could compromise the health of resident fish populations, benthic communities and plankton, even where concentrations remain sub-lethal. Over time, such pressures may reduce biodiversity, alter species composition, and weaken key ecosystem services such as fisheries production and nutrient cycling. These ecological risks reinforce the need for stricter control of pharmaceutical effluents and integrated monitoring that includes biota and habitat quality alongside water chemistry. The findings thus provide actionable evidence to guide environmental policy and industrial waste-treatment strategies at both national and regional levels.

This study has some limitations that should be considered when interpreting the findings. The sampling was conducted during a single season at a limited number of sites, which may not have adequately represented temporal variations or finer spatial differences in heavy metal concentrations within the Turag and Shitalakshya rivers. Furthermore, the health risk assessment considered only ingestion and dermal exposure to surface water, omitting other significant pathways such as consumption of contaminated fish, crops irrigated with polluted water, or inhalation of resuspended particles which are likely to be important exposure routes in this region. The exposure models that were applied relied on fixed default parameters and population averages, which do not account for individual differences in behavior, physiology, or local water-use habits, and we did not perform a formal sensitivity or uncertainty analysis of these assumptions. Although pollution indices, correlation analysis, PCA, and HCA were used to identify patterns and groupings of metals, key physicochemical parameters including pH, electrical conductivity, dissolved oxygen (DO), total dissolved solids (TDS), biochemical oxygen demand (BOD), chemical oxygen demand (COD), and temperature were not incorporated into the multivariate analysis. As a result, the statistical evaluation remains largely descriptive rather than mechanistic. Future research should include seasonal sampling, an expanded site network, concurrent measurement of water quality variables, and more advanced source apportionment methods to better understand metal mobility, geochemical behavior, and pollution origins in river systems affected by pharmaceutical-industrial discharges.

Conclusion

This study provides an integrated assessment of heavy metal contamination and associated health risks in surface waters receiving pharmaceutical effluents in Gazipur and Narayanganj, Bangladesh. The combined evidence from pollution indices, health risk models and multivariate analyses demonstrates that several sites are critically impacted by heavy metals and that children are particularly vulnerable to both non-carcinogenic and carcinogenic risks. The grouping of key metals in the multivariate analysis is consistent with inputs from pharmaceutical-industrial activities and indicates inadequate control of wastewater discharges.

These findings have important implications for environmental management and public health. They highlight the need for stricter enforcement of effluent discharge standards, routine auditing of effluent treatment plants, and the adoption of cleaner production and advanced treatment technologies capable of reducing metal loads at source. Establishing a coordinated monitoring framework for industrial river corridors, with regular surveillance of heavy metals and core water-quality parameters, would support early detection of hotspots and more effective regulatory action. Future work should be built on this baseline by incorporating multi-season sampling, additional sites along the river network, and concurrent measurements of physico-chemical parameters and biota to better characterize metal mobility, ecological impacts and dietary exposure pathways. Scenario-based evaluations of different treatment and regulatory options would further assist policymakers in prioritizing interventions. Overall, the study underlines that effective management of pharmaceutical-industrial effluents is essential to safeguard both human health and riverine ecosystems in Bangladesh and similar rapidly industrializing settings.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: SN, AR; Writing—Original draft: AR, AA, LK, SM; Figure generation: AR, AA and LK; Laboratory analysis: MM; Writing—review and editing: SN, AR, SM, AA, LK. All authors revised the manuscript, approved the final submitted version, agreed to be personally accountable for their contributions, and will ensure that any concerns regarding integrity or accuracy are investigated and resolved.

Funding

The authors did not receive any funding for this work.

Data availability

Corresponding author will be ready to deliver the supplementary and raw data based on the query or requirement regarding the research.

Declarations

Competing interests

The authors declare no competing interests.