Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 May 14.
Published in final edited form as: J Water Health. 2026 Jan 30;24(2):160–175. doi: 10.2166/wh.2026.175

Case-Control Study of Water and Agricultural Exposures Among Women in CKDu-Endemic Sri Lanka

Marimar Contreras Nieves a, Santhushya Hewapathiranage b, F Quin Zabel c, Dhananjalee Mahalekam d, Xue Yu a, Chiran Weerakoon b, Manoj Bandara b, Isuru Bandara b, Ishan Maduwantha b, Dulanjalee Hearth b, Ishara Athauda d, Bhagya Rathnayake d, Naduni Erandika b, Nilantha Pushpakuamra b, Eddie Lee a, Kyla Baron e, Vivek Bhalla a, Karen L Artiles f, Andrew Fire f, Stephen L Schensul g, Rohana Chandrajith d,*, Penny Vlahos c,*, Maria Montez-Rath a,*, Shuchi Anand a,*,#, Nishantha Nanayakkara b,*
PMCID: PMC13170613  NIHMSID: NIHMS2171839  PMID: 41764389

Abstract

Chronic kidney disease of unknown etiology (CKDu) disproportionately affects men in agricultural regions; few studies parse exposures among affected and unaffected women. We conducted a case-control study among women in Sri Lanka, identifying 15 cases who experienced progressive kidney disease in a prospective study in a CKDu-endemic region. We selected age-matched controls from the same endemic and neighboring non-endemic region. We then assessed differences in self-reported exposures, and analyzed water samples for inorganic and organic chemicals. Women who reported heat exhaustion (OR=19.3; p=0.017) and pesticide handling (OR=12.6; p=0.011) were more likely cases. Women who used natural fertilizer in home garden were more likely controls (OR=0.05; p=0.008). Cases’ water had higher fluoride, sulfate, calcium, and vanadium, and more often contained pesticides at >1 ppb, particularly diazinon, compared with controls from endemic and non-endemic regions. Our study supports prior data linking heat exposure to CKDu and identifies differences in home gardening, pesticide use, and water content among CKDu-affected women.

Keywords: agriculture, CKDu, fluoride, groundwater contamination, pesticides

1. Introduction

Chronic kidney disease of unknown etiology (CKDu) is the leading cause of tubulointerstitial kidney disease globally, disproportionately affecting young-to-middle-aged residents of agricultural communities in specific locations. Since prevalence is 2- to 3- fold higher among men than in women, extant data exploring risk factors focus largely on men (Gonzalez-Quiroz et al., 2018, Keogh et al., 2022, Glaser et al., 2020). These data report homogenous exposures among affected and unaffected men (Gonzalez-Quiroz et al., 2023, Hewavitharana et al., 2023a). In our study regions in Sri Lanka, for example, we reported that a wide range of potential exposures exist but very few are distinct between men with existing CKDu and rapid kidney function decline, versus those who have stabilized (Vlahos et al., 2021, Hewavitharana et al., 2023b). The majority of men reported farm work (96%), prior well water use (90%), and tobacco use (97%) (Hewavitharana et al., 2023b).

Epidemiologic principles point to the importance of outlier analysis, as the disease-conferring exposure may be more easily distinguished among outliers (Centers for Disease Control and Prevention, 2023). Classic examples date to the original cholera investigations of John Snow, during which he identified a patient who lived outside the disease outbreak area. This outlier patient was having water delivered from the water pump in the outbreak area, a critical datapoint in support of Snow’s hypothesis that contaminated water was the route of exposure (Reid, 2022). Given that women may be considered collective outliers in the context of CKDu, in whom heterogeneity in occupational or environmental exposures may be more informative than the largely homogenous exposures observed among studies focused on men, we conducted a case–control study of women living in two agricultural communities in Sri Lanka. (Janoudi et al., 2024).

We hypothesized two waterborne pathways through which well-water exposures may contribute to CKDu risk: 1) agriculture-related contaminants are more likely to concentrate in CKDu-endemic regions than in non-endemic regions of Sri Lanka, and 2) the inorganic composition, particularly the higher fluoride concentration or the water hardness in endemic regions, may enhance physiologic uptake of these contaminants. Further we hypothesize that heat stress may amplify nephrotoxic effects of contaminated water.

Using data from two prospective studies sampling kidney function and household water, we identified cases of women with established CKD who were experiencing rapid kidney function decline residing in Wilgamuwa, a region identified by the Ministry of Health as high prevalence for CKDu (Ranasinghe et al., 2019). We selected two age-matched control groups without CKD: one from Wilgamuwa, and another from a neighboring low-prevalence area (Hasalaka). We tested the association of self-reported residential history, family history, water use, and occupational history with the likelihood of being a case versus control. In addition, we sampled participants’ water and characterized differences in the inorganic and organic (agrochemical) content among cases versus controls.

2. Methods

This study collates data from two ongoing prospective cohorts. Cases are drawn from the Kidney Protection Project (KiPP) and controls from the Surveillance of Persons at Risk for Kidney Disease (SPARK) study. The KiPP study received Ethics approval from the University of Connecticut and National Teaching Hospital in Kandy, Sri Lanka. Stanford University approved follow-up activities. The SPARK study received Ethics approval from Stanford University and National Hospital in Kandy, Sri Lanka.

2.1. The KiPP Study: Selection of Cases

The KiPP study, a prospective cohort initiated in 2017, recruited 292 participants (221 men, 71 women) with moderate to advanced kidney disease, as measured by estimated glomerular filtration rate (eGFR) 20–59 ml/min/1.73 m2 from the Wilgamuwa Divisional Secretariat (Hewavitharana et al., 2023a). With a population of ~30,000 persons, this rural Divisional Secretariat in Sri Lanka is located in the lowland, dry zone of the Central Province, with rice as its primary crop (Fig. 1).

Fig. 1.

Fig. 1.

Open in a new tab

Map of Sri Lanka showing the CKDu-endemic region of Wilgamuwa (cases and controls) and the non-endemic area of Hasalaka (controls).

The study’s detailed methods are described elsewhere (Vlahos et al., 2019). Briefly, this cohort provided a baseline questionnaire, clinical measurements, and well water samples, and has had kidney function measurements (serum creatinine) over time to determine rate of loss of kidney function. For the first two years of the study, we measured serum creatinine every six months, and as of 2020, we have measured serum creatinine and/or obtained vital status annually.

Within this cohort of persons with established kidney disease, we identified women (cases) who, by June 2024, had experienced a ≥30% sustained decline in eGFR from baseline, determined using two consecutive measurements, or had reached eGFR <15 ml/min/1.73 m2 during the study period. A ≥30% sustained decline in eGFR has been associated with 5 to 7-fold higher relative rate of ESKD and 2-fold higher relative rate for mortality, compared with no change in eGFR over an equivalent time period (Coresh et al., 2014). This case definition facilitated selection of persons with a more definitive disease phenotype. By focusing on “decliners,” we aimed to reduce misclassification bias, and improve specificity. In addition, this definition increased the likelihood that baseline exposures at enrollment may influence observed change in kidney function during follow up.

2.2. The SPARK Study: Selection of Controls

The SPARK study is an ongoing prospective cohort of individuals residing in either a CKDu endemic area (Wilgamuwa) or adjacent non-endemic areas, including Hasalaka, where rice is also the primary crop. Participants in the study met the inclusion criteria if they were 18–60 years old and had serum creatinine level below 1.3 mg/dL (115 μmol/L) for men and 1.1 mg/dL (97 μmol/L) for women, no more than trace protein or blood on urine dipstick, and no self-reported diabetes. Similar to the KiPP cohort, these participants underwent a comprehensive questionnaire on demographics, medical and residential history, behavioral factors, laboratory evaluation, and water sampling at baseline. As of June 2024, we had enrolled 444 participants (220 men, 224 women).

From this cohort, we selected two distinct sets of age-matched (±10 years) female controls, one residing in Wilgamuwa and the other in Hasalaka. If an individual could serve as a control for more than one case, we prioritized matching cases with the least number of available controls. If we could not identify an age-matched control, we targeted recruitment for that age group in the next SPARK field visit. When multiple potential matches existed, we randomly selected from the pool.

Baseline questionnaire data from KiPP and SPARK were largely concordant, but when there were missing data for a specific question in either study (e.g., number of years lived in Wilgamuwa or number of family members), we instituted additional questions as part of a supplement questionnaire during routine scheduled visits.

2.3. Water sampling

Although a majority of women residing in Wilgamuwa are now drinking reverse osmosis (RO) water, well water has been their historic primary drinking water source (de Silva and Albert, 2021). We obtained a single timepoint sample of the participants’ well water at the time of recruitment into the study. In Hasalaka, drinking water sources include tube wells and pipe-borne water, and we sampled these as applicable to the participant. We sampled wells associated with CKDu case households between February and August 2018. We sampled wells or other drinking water sources associated with CKDu control households between January and May 2023. We collected well water by pulling up the well bucket and pouring into a 1-L amber glass bottle. Then, we filtered the water through a 0.45 mm glass fiberglass filter and run it through a C-18 solid phase extraction (SPE) column within 24 hours of collection, storing the columns at −20°C until elution.

2.4. Inorganic water analysis

Inorganic analysis of water samples characterized pH, electric conductivity (EC), fluoride, phosphate, and major/trace metals according to methods reported in Balasooriya et al. (Balasooriya et al., 2020). For major ion analyses, we filtered an aliquot of 100 mL through 0.45-μm pore size disposable nylon syringe filter into falcon Nalgene tubes. For cation analyses, we filtered 15 mL of water into a separate falcon tube and acidified with ultrapure nitric acid. We used ion chromatography to measure chloride (Cl−), fluoride (F−), nitrate (NO3−), phosphate (PO43−), and sulfate (SO42−) (Thermo Dionex ICS-1100), with relative standard deviations below 5%. We used an atomic absorption spectrometer (Varian 240FS) to determine the major cations, and inductively coupled plasma mass spectrometry (Thermo Fisher ICapQ) to measure trace elements.

2.5. Organic water analysis

We extracted sample cartridges with acetonitrile (99.6% for analysis, ACROS Organic) and spiked them with surrogate standards acenaphthene-d10, 1,4-dichlorobenzene-d4, and chrysene-d12, each at 10 mg/L, before concentrating them down to 1-mL volume under N2 flow. We performed analysis for case water samples with an Agilent 7890A gas chromatograph (GC) coupled with an Agilent 5975C mass spectrometer (MS) (Agilent, USA), and performed analysis for control water samples with a Thermo Scientific GC Trace 1600 coupled with ISQ 7610 single quadrupole MS (Thermo Scientific, USA). We paired both mass spectrometers with an Rtx-5MS column with Integra-Guard and analyzed samples in splitless mode (Restek, USA). We injected samples at a 1 μL volume. We set the initial temperature to 65°C, held it for 4 minutes, ramped it by 10°C/min until reaching 280°C, held for 10 minutes, and applied a 2-minute post-run at 300°C. We set the limit of detection at three times the baseline noise threshold.

We performed full non-targeted scans for all samples to identify compounds for targeted analysis. Based on the agrochemicals known to be applied in the area and results of the non-targeted scan, we compiled three standard mixes of relevant pesticides, amounting to 33 pesticides in total, and used them as standard mixes. We then re-analyzed the samples in selected ion monitoring (SIM) mode and quantified pesticide concentrations based on standard curves. We then adjusted concentrations based on the percent yield of the surrogate standards introduced during extraction. Because all pesticides were quantified based on external calibration standards, concentrations are comparable between case and control batches.

Glyphosate is highly polar and has low volatility. Therefore, derivatization steps are required to make glyphosate amenable to GC-MS analysis (Denžić Lugomer et al., 2024). Notably, we did not take such derivatization steps in this analysis. Thus, while the tested samples may contain glyphosate, the method used here does not target this molecule for quantification.

2.6. Statistical Analyses

We performed exploratory analyses on potential exposures across several categories, including ongoing exposures since childhood, occupational exposures, water usage history, and behavioral factors (Table 1). We summarized and compared baseline characteristics by case versus control status using the original data (complete case analysis) and exact logistic regression (Stata command exlogistic) (Cox, 1989). For each exposure, we then performed multiple imputation and estimated odds ratios (OR) for being a case versus control, using logistic regression with robust standard errors. Each exposure was evaluated in a separate model accounting for age and education. We selected logistic regression over conditional logistic regression to allow for age adjustment despite approximate age matching, and used robust standard errors to account for the clustering of observations within matched pairs. Given the exploratory nature of this study and the small sample size, we highlight ORs with a p-value<0.1.

Table 1:

Baseline Characteristics of Endemic (Wilgamuwa) Cases and Controls and Non-Endemic (Hasalaka) Controls$

Endemic Cases Endemic Controls Non-Endemic Controls
Variable Mean ± SD
or N (%)
N=15# 		Mean ± SD
or N (%)
N=15# 		Mean ± SD
or N (%)
N=15#
Demographics
 Age 53.3 ± 8.1 49.9 ± 6.0 51.0 ± 8.1
 Years of education 4.4 ± 3.3 7.5 ± 4.5 8.5 ± 4.2**
 eGFR 36.6 ± 29.5 90.8 ± 15.6 99.8 ± 13.0
Family History
 Any first generation family member with kidney disease 7 (47%) 6 (40%) 0 (0%)**
 Any first generation family member died of kidney disease 5 (33%) 3 (20%) 0 (0%)*
Exposures since childhood
 Born in Wilgamuwa 9 (60%) 10 (67%) 0 (0%)
 Years in Wilgamuwa 39.5 ± 17.0 50.0 ± 5.9
Water usage history
 Distance to closest reservoir in kilometers 1.7 ± 0.8 1.7 ± 0.4
 History of well water use 12 (80%) 15 (100%) 12 (80%)
  Well water as birthplace primary drinking water 10 (83%) 11 (73%) 8 (53%)
  Years of well water use 44.9 ± 3.18 41.3 ± 10.1 24.0 ± 17.1***
  Owned the household well (vs shared with others) 9 (75%) 15 (100%) 8 (53%)
  Self-reported well depth (meters) 13.1 ± 3.09 9.5 ± 2.6** 13.9 ± 7.0
  Bucket to access well 5 (42%) 6 (40%) 3 (20%)
  Pipes to access well 0 (0%) 2 (13%) 0 (0%)
  Motor pump to access well 12 (100%) 13 (87%) 7 (47%)**
  Other source to access well 0 (0%) 0 (0%) 1 (7%)
 Current use of well water 4 (27%) 3 (20%) 3 (20%)
 History of reverse osmosis water use 8 (53%) 15 (100%) 1 (7%)*
 Current use reverse osmosis water 8 (53%) 12 (80%) 1 (7%)*
 Current use of paddy water for drinking 1 (7%) 1 (7%) 1 (7%)
Occupational exposures
 Ever experienced heat exhaustion symptoms 7 (47 %) 0 (0%)** 1 (7%)*
 Farmed in paddy fields 9 (75%)a 14 (93%) 14 (93%)
  Age started paddy farming (years) 18.7 ± 4.1 19.1 ± 4.1 19.4 ± 5.7
  Paddy farming tasks:
   Clearing/preparing field 5 (42%) 9 (60%) 9 (60%)
   Burning field 0 (0%) 1 (7%) 1 (7%)
   Plowing 4 (33%) 5 (33%) 6 (40%)
   Seeding/ transplanting 8 (67%) 14 (93%) 12 (80%)
   Weeding 7 (58%) 10 (67%) 12 (80%)
   Harvesting 7 (58%) 3 (20%) 9 (60%)
   Threshing 4 (33%) 1 (7%) 12 (80%)*
   Drying 8 (67%) 13 (87%) 14 (93%)
   Transport 0 (0 %) 0 (0%) 1 (7%)
 Household practiced field burning 2 (17%)a 1 (7 %) 1 (7%)
 Any pesticide handling 11 (73 %) 3 (20%)** 7 (47%)
 Glyphosate herbicide application 0 (0%)a 1 (7%) 0 (0.00%)
 Owns a home garden 12 (92%)b 15 (100%) 13 (87%)
 Home garden products a
  Natural fertilizer 2 (17 %) 12 (80%)** 10 (7%)*
  Store-bought fertilizer 9 (75%) 3 (20%)* 8 (53%)
   Insecticides 8 (67%) 2 (13%)* 7 (47%)
   Herbicides 11 (92 %) 0 (0%)*** 0 (0%)***
   Fungicides 0 (0%) 0 (0%) 0 (0%)
   Fumigants 0 (0%) 0 (0%) 0 (0%)
   Used well water to garden 0 (0%) 14 (93%)*** 0 (0%)
   Used reservoir water to garden 1 (8%) 1 (7%) 1 (7%)
Behavioral factors
 Fresh water (local) fish consumption 7 (58%)a 14 (93%) 10 (67%)
 Ever used tobacco 2 (13%) 0 (0%) 1 (7%)
  Current tobacco use 2 (100%) 0 (0%)* 1 (7%)*
 Ever used alcohol 0 (0%) 0 (0%) 0 (0%)
Open in a new tab
#

N=15 unless noted otherwise in the cell.

$

P-value from exact logistic regression on original data (unadjusted).

*

P-value<0.05.

**

P-value<0.01.

***

P-value<0.001

Superscript letters Indicate number of responses:

a

N = 12,

b

N = 13

Each multiple imputation model assumed the data to be missing at random given age, education and the matched pair and generated 10 imputed datasets. For continuous exposures, we used predictive mean matching with 5 nearest neighbors while, for categorical exposures, we used logistic regression. We combined the results using Rubin’s rules (Montez-Rath et al., 2014, Little, 2002).

Imputation quality was evaluated by comparing the distribution of observed and imputed values using summary statistics (means and standard deviations) across imputations. For comparison, we also estimated complete-case (“observed”) ORs. Complete-case and multiple-imputation analyses produced similar results; we selected multiple imputation as the primary approach because it relies on the less restrictive missing-at-random assumption rather than missing completely at random.

Inorganic data analysis included the comparison between cases and each control group using Kolmogorov-Smirnov tests (Dodge, 2008). We further tested the linear trend of the associations, assuming that if an order exists in the data, it is 1) endemic (Wilgamuwa) cases, 2) endemic (Wilgamuwa) controls, and 3) non-endemic (Hasalaka) controls. To account for multiple comparisons across 90 tests, we applied a Bonferroni correction, setting our significance level for the inorganic data analysis at α = 0.00056 (0.05/90). For water organic (agrochemical) analysis, we report the number of detections >1 part per billion (μg/L) by cases versus control status. For water analyses we used the water source as the unit of observation. When multiple control participants reported sharing the same household water source, the corresponding inorganic and organic measurements were counted once per unique water source rather than per person to avoid pseudoreplication and artificial inflation of sample size.

All statistical analyses were performed using Stata, version 18.0 (StataCorp LLC) (StataCorp, 2024).

3. Results

There were 15 cases with declining kidney function, and 30 matched controls. Mean age of cases was 53 (SD 8) years, CKDu endemic area controls was mean age 50 (SD 6) years and non-endemic area controls was 51 (SD 8) years (Table 1). Cases had fewer years of education than endemic and non-endemic area controls. Family history of kidney disease was present in cases (47%) and endemic area controls (40%); but was absent among non-endemic area controls (Table 1).

Among all groups, a majority of women engaged in farming (9 (75%)), starting between age 18–19 years, and used well water at some point in their lives (12 (80%)), although endemic area women reported longer lifetime mean duration of well water use (Table 1). Reverse osmosis (RO) water use at the time of study recruitment was present exclusively in the endemic area. Self-reported glyphosate use and household field burning practices were rare across all groups. Heat exhaustion symptoms were present among case women (7(47%)) but not among endemic controls. Only cases reported use of herbicides in the home garden (11 (92%)); none of the endemic or non-endemic area controls reported herbicide use. In addition, there was minimal self-reported use of tobacco and alcohol (Table 1).

3.1. Differences between case and CKDu endemic-area resident control women

In age and education adjusted comparisons between endemic-area cases and controls, we did not identify differences in length of residence within the endemic region. Women reporting use of RO water at time of recruitment were less likely to be cases (Table 2a). Among occupational tasks, women who participated in harvesting and handled pesticides were more likely to be cases. Women who reported using home garden pesticide and insecticide use were more likely to be cases, whereas women reporting use of natural fertilizer were less likely to be cases. Women reporting no fresh water fish consumption were more likely to be cases.

Table 2a:

Comparison between Endemic Cases and Controls (Wilgamuwa)$

Variable# Odds of being a case (OR), relative to endemic area controls p-value
Self-reported well depth (meters) 2.27 0.025
Current reverse osmosis water use 0.21 0.073
Harvesting during paddy farming 8.11 0.056
Any pesticide handling 12.6 0.011
Home garden products
 Natural fertilizer 0.05 0.008
 Store-bought fertilizer 37.8 0.003
 Insecticides 16.3 0.037
Freshwater (local) fish consumption 0.06 0.048
Open in a new tab
$

From logistic regression models adjusted for age and education with robust standard errors considering the matched pair.

#

Variables with p-values < 0.10. See Table S1 for full list of variables evaluated. OR = Odds ratio.

3.2. Differences between case and non-endemic area resident control women

In comparisons between non-endemic area controls and endemic area cases, women who reported experiencing heat exhaustion, owning the household well, and owning a home garden were more likely to be cases (Table 2b). Consistent with differences between endemic resident cases and controls, women who reported using natural fertilizer in the home garden were more likely controls.

Table 2b:

Comparison between Endemic (Wilgamuwa) Cases and Non-Endemic (Hasalaka) Controls$

Variable# Odds of being a case (OR), relative to non-endemic controls p-value
Years of education 0.76 0.004
Owned the household well (vs shared with others) 18.3 0.037
Bucket to access well 8.48 0.045
History of reverse osmosis water use 34.8 0.005
Current reverse osmosis water use 34.8 0.005
Ever experienced heat exhaustion symptoms 19.3 0.017
Threshing during paddy farming 0.08 0.051
Drying during paddy farming 0.16 0.094
Owns a home garden 15.36 0.029
Home garden products
 Natural fertilizer 0.16 0.028
Open in a new tab
$

From logistic regression models adjusted for age and education with robust standard errors considering the matched pair.

#

Variables with p-values < 0.10. See Table S1 for full list of variables evaluated. OR = Odds ratio.

Additional results with adjusted odds ratios from logistic regression analyses are presented in Table S1.

3.3. Inorganic water analyses

All measured inorganic constituents in drinking water samples from cases and endemic and non-endemic controls were within safety limits defined by the World Health Organization (WHO) and U.S. Environmental Protection Agency (EPA). Median pH was slightly below the EPA-recommended range across all groups (Table S2).

Compared with endemic-area controls, case household water samples showed higher concentrations of suspected nephrotoxicants or agrochemical uptake enhancers, i.e., fluoride and vanadium. Cases’ water samples also had elevated levels of hardness-related and geochemically common ions: calcium, sodium, potassium, sulfate, and nitrate. In addition, other biologically active trace elements, including copper, zinc, and selenium, had higher levels in case wells. In contrast, several metals typically considered nephrotoxicants, including arsenic, cadmium, and lead, were lower in cases than in endemic-area controls. (Table S2, Fig. 2)

Fig. 2.

Fig. 2.

Open in a new tab

Boxplots showing the distribution of selected inorganic chemicals in drinking water from CKDu-endemic Wilgamuwa (cases and controls) and non-endemic Hasalaka (controls). WG = Wilgamuwa; HS = Hasalaka. Number of samples and trend p-value (n Cases WG, n Controls WG, n Controls HS; p-value): Phosphate (15,11,5; 0.2223), Fluoride (15,14,5; 0.0001), Nitrate (7,13,5; 0.6715), Sulfate (15,14,5; 0.0043), Calcium (15,14,5; 0.0100), Vanadium (15,15,5; 0.0005), Arsenic (15,15,5; 0.1830), Cadmium (15,15,5; 0.0020), Lead (15,15,5; 0.0967). Additional inorganic constituents not shown are fully reported in Table S2.

Relative to non-endemic-area controls, case water samples showed a similar trend with higher concentrations of suspected nephrotoxicants or agrochemical uptake enhancers (fluoride, and vanadium), hardness-related ions (calcium, magnesium, sodium, potassium, and sulfate), and other biologically active trace elements (zinc, aluminum, and selenium). In contrast, cadmium, beryllium, nickel, and lead, were lower in cases, while arsenic levels were similar between groups (Table S2, Fig. 2).

3.4. Organic water analyses

In non-targeted analyses of case wells, we identified 13 agrochemicals which were further quantified for limit of detection (>1 ppb) (Table 3). Overall frequencies of detection were higher in case wells, relative to endemic area control wells. Notably, none of the non-endemic area control samples had detections >1 ppb. Three compounds (specifically Endosulfan II, diazinon, and p, p’-DDE) were detected more frequently in case samples than in controls.

Table 3:

Detection and Concentration of Agrochemicals in Well Water Samples Among Endemic (Wilgamuwa) Cases and Controls and Non-Endemic (Hasalaka) Controls

Compound WHO Pesticide Hazard Ranking** Cases Detections* > 1 ppb N (%) Cases Concentration# [Median (Range)] Endemic Controls* Detections > 1 ppb N (%) Endemic Controls Concentration# [Median (Range)] ppb Non-Endemic Controls* Detections > 1 ppb N (%)
Organochlorine insecticides
 Lindane 2 0 (0.0) 0 1 (6.7) 7.086 0 (0.0)
 Heptachlor -- 0 (0.0) 0 1 (6.7) 6.668 0 (0.0)
 Propanil 2 1 (6.7) 3.231 0 (0.0) 0 0 (0.0)
 Endosulfan II 2 3 (20.0) 2.592 (2.267 – 2.621) 0 (0.0) 0 0 (0.0)
 o,p’-DDT 2 0 (0.0) 0 0 (0.0) 0 0 (0.0)
 p,p’-DDE -- 3 (20.0) 2.232 (1.292 – 2.544) 1 (6.7) 1.565 0 (0.0)
 Endrin Ketone -- 0 (0.0) 0 0 (0.0) 0 0 (0.0)
Organophosphate insecticides
 Diazinon 2 8 (53.3) 1.8595 (1.333 – 7.505) 1 (0.0) 5.651 0 (0.0)
Chloroacetanilide herbicides
 Clomazone 2 0 (0.0) 0 2 (13.3) 4.931.5 (1.071 – 8.792) 0 (0.0)
 Pretilachlor -- 1 (6.7) 1.952 1 (6.7) 1.007 0 (0.0)
 Propachlor 2 0 (0.0) 0 0 (0.0) 0 0 (0.0)
 Metolachlor 3 0 (0.0) 0 0 (0.0) 0 0 (0.0)
Triazinone herbicide
 Metribuzin 2 0 (0.0) 0 0 (0.0) 0 0 (0.0)
Open in a new tab
*

Among Wilgamuwa cases and controls, water samples were obtained from 15 wells. In Hasalaka controls, samples from 5 wells were collected, representing 7 total individuals, with 1 well corresponding to 3 individuals who reported sharing the same well.

#

Median and range of concentrations among participant wells with detections >1 ppb.

All samples were analyzed using gas chromatography–mass spectrometry (GC-MS) with targeted pesticide quantification.

**

World Health Organization (WHO) Pesticide Hazard Score Ranking is based on known pesticide toxicity. A score of 2 indicates that a pesticide is “moderately hazardous” while a score of 3 indicates that a pesticide is “slightly hazardous”. Unranked pesticides may be unlikely to cause harm or toxicity, or may have been deemed obsolete and unlikely to be encountered by the WHO(World Health Organization, 2019).

4. Discussion

In this case control study comparing the characteristics of women with progressive kidney disease living in a CKDu endemic area with similarly aged healthy women living in endemic and non-endemic areas, we find that women with progressive kidney disease were more likely to report heat stress symptoms and handling agrochemicals. Home garden practices also distinguished cases and controls: women relying on natural fertilizers and avoiding herbicides were more likely healthy controls. Although inorganic chemical levels in household water samples were below thresholds of concern, samples from cases indicated modestly higher levels of fluoride and vanadium, and more frequent detections of agrochemicals, relative to controls from both areas.

We also report some notable ‘negative’ findings or associations, which could be equally consequential in informing future investigations, since we selected outlier cases with progressive kidney disease. Specifically, self-reported family history of kidney disease was common in the endemic area, but not different between cases and controls, while it was absent in the non-endemic area. There was also low self-reported prevalence of field burning—a hypothesized route for silica nanoparticle exposure—and use of tobacco or alcohol in both endemic and non-endemic area women. Finally, cadmium, lead, and arsenic—known nephrotoxicants—were low in water samples across all groups and lowest in case samples.

The association between heat exhaustion symptoms and CKDu case status is consistent with prior studies from Central America. In a seven-year cohort of residents of a Mesoamerican Nephropathy (MEN) hotspot in Nicaragua, Gonzalez-Quiroz et al. found that cumulative years of sugarcane work and self-reported symptoms of excess sun exposure were associated with incident CKD (Gonzalez-Quiroz et al., 2023). Similarly, Kupferman et al. identified a high incidence of acute kidney injury during the harvest season among sugarcane workers in Nicaragua, with nearly half of those affected developing CKD within one year (Kupferman et al., 2018).

Nearly half of our cases and endemic area controls reported a first-generation family member with kidney disease. Although we do not have genotype data for our sample, whole genome analyses indicate shared genetic ancestry across the major ethnic populations in Sri Lanka(Urban Aragon et al., 2025). In our study, women from the nearby non-endemic area, who live within the same regional context but experience different environmental exposures, reported substantially less family history of kidney disease. This pattern is consistent with a possible distinct environmental or environmental-genetic interaction in the endemic area (Nanayakkara et al., 2015, Bandara et al., 2022). Much of the CKDu-endemic Dry Zone was repopulated during the late 20th century through major state-led agricultural resettlement programs, particularly the Mahaweli Development Project (Wimalawansa, 2014). These movements brought genetically similar populations into different environmental exposures, providing a natural context in which environmental exposures may modulate genetically mediated CKDu risk (Nanayakkara et al., 2015, Bandara et al., 2022).

The association between pesticide handling and CKDu case status observed in our study is noteworthy amid the ongoing debate regarding the role of pesticides in CKDu. In a systematic review, Valcke et al. found limited evidence linking pesticide exposure to CKDu due to poor exposure assessment and confounding factors (Valcke et al., 2017). Smpokou et al. tested pesticide metabolites in urine among their community-based cohort in Nicaragua but found no association with kidney function decline (Smpokou et al., 2019). In contrast, Jayasumana et al. reported a significantly increased risk of CKDu among Sri Lankan paddy farmers who sprayed glyphosate (Jayasumana et al., 2015). Our study results suggest self-reported history of ever handling pesticides as a potentially relevant occupational exposure, particularly among women with distinct exposure pathways (i.e., through the home garden). Only cases in our study reported use of herbicides in the home garden, Additionally, controls in our study were more likely to report using natural fertilizers, suggesting that these practices could play a protective role.

The use of herbicides or other synthetic inputs in the home garden could contaminate the home well. Several experts hypothesize that drinking water contaminated with agrochemicals in combination with water hardness and co-occurring trace elements contribute to CKDu in Sri Lanka (Ulrich et al., 2023, Chandrajith et al., 2024, Jayasumana et al., 2014). In our study, we incorporated analysis of drinking water sources to evaluate both inorganic constituents and agrochemical residues, and observed patterns that seem to align with these proposed exposure pathways, although causal inference is beyond the scope of this case–control study.

In our analysis of agrochemicals, Diazinon, Endosulfan II, and p,p’-DDE, were detected more frequently in case wells. Diazinon, an organophosphate insecticide, has been shown in animal studies to cause oxidative stress-mediated kidney damage through depletion of antioxidant enzymes and increases in lipid peroxidation, leading to histopathological evidence of renal injury (Shah and Iqbal, 2010). Endosulfan II, a component of the organochlorine pesticide endosulfan (banned in the US as of 2010 (US Environmental Protection Agency, 2016)), has been associated with acute kidney injury in both human poisoning cases and experimental animal models, with findings including tubular necrosis, glomerular damage, and disrupted antioxidant activity (Blanco-Coronado et al., 1992, Lo et al., 1995, Caglar et al., 2003). p,p’-DDE, a persistent metabolite of DDT, has been linked to accelerated decline in glomerular filtration rate in a prospective cohort study (Jayasinghe et al., 2018), although animal studies have shown mixed evidence of kidney effects (Deichmann et al., 1967, Makita et al., 2003). p,p’-DDE likely represents a legacy compound and may instead serve as a proxy for higher overall chemical exposure among persons in these households.

Our analysis of inorganic constituents in drinking water revealed that case wells had relatively higher concentrations of fluoride, sulfate, calcium, magnesium, and vanadium, and lower concentrations of arsenic, cadmium, and lead, compared to endemic and non-endemic controls. Some of the inorganic chemicals that we detected (e.g. sodium, calcium, magnesium, and potassium) are common in Sri Lankan Dry Zone groundwater and do not have established nephrotoxic effects at the concentrations detected.

Fluoride concentrations in our samples averaged 0.5 mg/L, below the WHO maximum of 1.5 mg/L and the Sri Lankan regulatory threshold of 1.0 mg/L. However, they meet the more conservative 0.5 mg/L benchmark proposed by multiple Sri Lankan researchers as a safer upper limit for tropical countries, where higher water intake and concurrent exposures may amplify nephrotoxic effects (Warnakulasuriya et al., 1992). Notably, we do not observe clinical signs typically associated with chronic fluoride toxicity in our study population, such as dental or skeletal fluorosis (Srivastava and Flora, 2020). However, there could nonetheless be a potential for synergistic toxicity with another compound or set of compounds in setting of relatively higher fluoride exposure among definitive cases. Prior studies have also consistently reported significantly higher fluoride concentrations in CKDu-endemic areas (Gobalarajah et al., 2020, Nanayakkara et al., 2020, Chandrajith et al., 2024).

Some have proposed a potential for synergistic toxicity of fluoride when combined with cations—particularly calcium and magnesium—while others have proposed synergistic toxicity when combined with agrochemicals (Liyanage et al., 2022, Chandrajith et al., 2024, Jayasumana et al., 2014). Fluoride handling by the kidney is urinary pH dependent, with higher reabsorption occurring in the kidney tubules at lower urinary pH (Whitford et al., 1976). A lower urinary pH may also enhance toxicity of agrochemicals including organophosphates such as Diazinon which are optimally effective at acidic pH (Dharmaratne, 2019, Ulrich et al., 2023). We emphasize that these mechanisms remain speculative and in vivo models are required to evaluate any potential synergistic effects (Jayasumana et al., 2014).

Vanadium, a naturally occurring trace element found in soil and groundwater, has recently emerged as a potential nephrotoxicant of concern (Ucibior et al., 2014). A recent study from Tlaxcala, Mexico—an area with elevated CKD rates in young adults—demonstrated a strong association between urinary vanadium concentrations and markers of tubular dysfunction, including albuminuria, NGAL, and KIM-1, even among adolescents with normal kidney function (Rojas-Lima et al., 2025).

Arsenic, cadmium, and lead have been studied as possible contributors to CKDu, since these compounds have established nephrotoxic effects. In our study, levels of all three were well below safety limits and showed no consistent elevation among cases, findings consistent with prior studies from endemic regions in Sri Lanka and Central America that largely discount them as primary etiologic agents (Chandrajith et al., 2024, Nanayakkara et al., 2019, Rango et al., 2015). Of note, chemical and trace-element exposures represent only one set of potential water-borne pathways contributing to CKDu, and microbial mechanisms, which were not evaluated in this study, remain an important area for future investigation.

Our study has several strengths. Our case definition was specific: women not only had established CKD, but also experienced decline in kidney function during prospective follow-up. Thus, their baseline reported exposures could be more likely implicated as potentially contributory to the observed loss of kidney function during the study period. We matched these cases to healthy women in both endemic and non-endemic (but demographically and occupationally) similar areas, allowing us to isolate disease-specific factors from community-pervasive factors. The use of detailed questionnaire data provided insights into lifetime occupational and environmental exposures, which are often difficult to capture in large-scale epidemiologic studies, and cannot be reliably measured using blood or urine samples. By focusing on women—an understudied population in CKDu research—our study aims to differentiate exposure pathways. Importantly, we incorporated environmental sampling of household drinking water, allowing us to objectively assess exposure to both inorganic elements and agrochemicals, with the latter analyses conducted with a comprehensive and untargeted approach. The concordant findings between our questionnaire-based and objective exposure data strengthens the internal validity of our findings.

Our study also has limitations. The small sample size limits its statistical power. However, the sample size reflects a deliberate hypothesis-driven sampling strategy in which we selected women based on prospectively observed kidney function decline, ensuring that baseline exposures preceded disease progression. Additionally, the case-control design, while valuable for identifying associations, does not allow us to establish causality between exposures and CKDu. Reliance on self-reported data for exposures such as heat stress symptoms, pesticide use, and water source history introduces potential recall and reporting bias, particularly if women with established kidney disease were more likely to reflect on or recall past occupational or household exposures. Such differential misclassification could exaggerate observed associations for these exposures. However, currently self-report remains the primary approach to estimate life-time exposures, since validated methods for cumulative (lifetime) heat stress or pesticide exposure do not exist.

Limitations also apply to the water analysis: we collected inorganic and organic samples at a single time point as part of a broader epidemiologic protocol focused on characterizing baseline exposures. The single time point sampling does not capture seasonal variability, and our current methods do not detect glyphosate. The time gap between case (2018) and control (2023) sampling could influence agrochemical detection because of changes in agricultural practices; interannual variation is unlikely for trace elements (Shiller, 1997). However, contemporaneous sampling of both regions in 2023 demonstrated higher number of agrochemical detections in the endemic region (6 versus 0 at > 1 ppb in endemic versus non endemic region, respectively). This is consistent with our hypothesis that well water in the endemic region is more likely contaminated with agrochemicals, compared with non-endemic region. All samples followed the same field collection and extraction protocol; however, the use of different instruments resulted in lower detection thresholds in the 2023 study and therefore a greater likelihood of detection in 2023. This analytical difference would bias against detection in 2018 and is not expected to alter our conclusions.

5. Conclusion

In summary, in our case-control analysis, we find that women with progressive loss of kidney function were more likely to self-report heat exposure and pesticide handling, with the home garden representing a potential route of exposure. Consistently we detected modestly more agrochemicals in their household wells, alongside relatively higher concentrations of some trace elements. Together, these findings identify patterns that are consistent with proposed groundwater-related exposure pathways in CKDu, but are associative and preliminary given the study design and sample size. Further prospective investigations are critical for testing our hypothesis-generating findings.

Supplementary Material

Supplemental material

Appendix A. Supplementary data

Acknowledgements

The authors thank the Sri Lanka Ministry of Health for its support and the women who participated in this study. We also acknowledge the laboratory teams for their assistance with water quality and agrochemical analyses. This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases through TL1/U2C 1DK133488-01A1, 3R01DK127138-04S1, and Ruth L. Kirschstein National Research Service Award (1F32DK142456-01A1) to Dr. Contreras-Nieves, and by 5R01DK127138 to Dr. Shuchi Anand.

Funding

This research was supported by the National Institute of Diabetes and Digestive and Kidney Diseases through TL1/U2C 1DK133488-01A1, 3R01DK127138-04S1, and Ruth L. Kirschstein National Research Service Award (1F32DK142456-01A1) to Dr. Contreras-Nieves, and by 5R01DK127138 to Dr. Shuchi Anand.

Footnotes

Conflict of interest

All authors declare no financial or non-financial competing interests.

Data availability

Data may be made available upon reasonable request following the completion of data sharing approvals from all three collaborating institutions: Stanford University, the University of Connecticut, and the National Teaching Hospital in Kandy, Sri Lanka.

References

  1. BALASOORIYA S, MUNASINGHE H, HERATH AT, DIYABALANAGE S, ILEPERUMA OA, MANTHRITHILAKE H, DANIEL C, AMANN K, ZWIENER C, BARTH JAC & CHANDRAJITH R 2020. Possible links between groundwater geochemistry and chronic kidney disease of unknown etiology (CKDu): an investigation from the Ginnoruwa region in Sri Lanka. Exposure and Health, 12, 823–834. [Google Scholar]
  2. BANDARA M, WEERAKOON C, HERATH H, HEWAPATHIRANAGE S, NADUNI ERANDIKA HG, HEWAVITHARANE P & NANAYAKKARA N 2022. Familial Clustering in Sri Lankan CKDu cases. [Google Scholar]
  3. BLANCO-CORONADO JL, REPETTO M, GINESTAL RJ, VICENTE JR, YELAMOS F & LARDELLI A 1992. Acute intoxication by endosulfan. J Toxicol Clin Toxicol, 30, 575–83. [DOI] [PubMed] [Google Scholar]
  4. CAGLAR Y, KAYA M, BELGE E & METE UO 2003. Ultrastructural evaluation of the effect of endosulfan on mice kidney. Histol Histopathol, 18, 703–8. [DOI] [PubMed] [Google Scholar]
  5. CENTERS FOR DISEASE CONTROL AND PREVENTION 2023. Lesson 6: Investigating an Outbreak, Section 2: Steps of an Outbreak Investigation. [Google Scholar]
  6. CHANDRAJITH R, NANAYAKKARA N, ZWIENER C, DANIEL C, AMANN K & BARTH JAC 2024. Geochemical Characteristics of Groundwater Consumed by Patients with Chronic Kidney Disease with Unknown Aetiology in the Crystalline Dry Zone Terrain of Sri Lanka. Exposure and Health, 16, 183–195. [Google Scholar]
  7. CORESH J, TURIN TC, MATSUSHITA K, SANG Y, BALLEW SH, APPEL LJ, ARIMA H, CHADBAN SJ, CIRILLO M, DJURDJEV O, GREEN JA, HEINE GH, INKER LA, IRIE F, ISHANI A, IX JH, KOVESDY CP, MARKS A, OHKUBO T, SHALEV V, SHANKAR A, WEN CP, DE JONG PE, ISEKI K, STENGEL B, GANSEVOORT RT & LEVEY AS 2014. Decline in estimated glomerular filtration rate and subsequent risk of end-stage renal disease and mortality. Jama, 311, 2518–2531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. COX DR, SNELL EJ 1989. Analysis of Binary Data, London, Chapman and Hall/CRC. [Google Scholar]
  9. DE SILVA MWA & ALBERT S 2021. Kidney Disease, Health, and Commodification of Drinking Water: An Anthropological Inquiry into the Introduction of Reverse Osmosis Water in the North Central Province of Sri Lanka. Human Organization, 80, 140–151. [Google Scholar]
  10. DEICHMANN WB, KEPLINGER M, SALA F & GLASS E 1967. Synergism among oral carcinogens. IV. The simultaneous feeding of four tumorigens to rats. Toxicol Appl Pharmacol, 11, 88–103. [DOI] [PubMed] [Google Scholar]
  11. DENŽIĆ LUGOMER M, BILANDŽIĆ N, PAVLIČEK D & NOVOSEL T 2024. Direct Determination of Glyphosate and Its Metabolites in Foods of Animal Origin by Liquid Chromatography–Tandem Mass Spectrometry. Foods, 13, 2451. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. DHARMARATNE RW 2019. Exploring the role of excess fluoride in chronic kidney disease: A review. Hum Exp Toxicol, 38, 269–279. [DOI] [PubMed] [Google Scholar]
  13. DODGE Y 2008. Kolmogorov–Smirnov Test. The Concise Encyclopedia of Statistics. New York, NY: Springer New York. [Google Scholar]
  14. GLASER J, HANSSON E, WEISS I, WESSELING C, JAKOBSSON K, EKSTRÖM U, APELQVIST J, LUCAS R, ARIAS MONGE E, PERAZA S, HOGSTEDT C & WEGMAN DH 2020. Preventing kidney injury among sugarcane workers: promising evidence from enhanced workplace interventions. Occup Environ Med, 77, 527–534. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. GOBALARAJAH K, SUBRAMANIAM P, JAYAWARDENA UA, RASIAH G, RAJENDRA S & PRABAGAR J 2020. Impact of water quality on Chronic Kidney Disease of unknown etiology (CKDu) in Thunukkai Division in Mullaitivu District, Sri Lanka. BMC Nephrol, 21, 507. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. GONZALEZ-QUIROZ M, HEGGESETH B, CAMACHO A, OOMATIA A, AL-RASHED AM, ZHANG Y, MCCREIGHT A, JEWELL N, ARAGON A, NITSCH D, PEARCE N & CAPLIN B 2023. Population-level detection of early loss of kidney function: 7-year follow-up of a young adult cohort at risk of Mesoamerican nephropathy. International Journal of Epidemiology, 53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. GONZALEZ-QUIROZ M, SMPOKOU ET, SILVERWOOD RJ, CAMACHO A, FABER D, GARCIA BR, OOMATIA A, HILL M, GLASER J, LE BLOND J, WESSELING C, ARAGON A, SMEETH L, PEARCE N, NITSCH D & CAPLIN B 2018. Decline in Kidney Function among Apparently Healthy Young Adults at Risk of Mesoamerican Nephropathy. J Am Soc Nephrol, 29, 2200–2212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. HEWAVITHARANA P, SCHENSUL S, LEE E, MONTEZ-RATH M, SENARATHNE S, LIU S, HAROLD K, HEWAPATHIRANAGE S, ERANDIKA N, ABEYSUNDARA HTK, YU X, BHALLA V, FIRE A, LEVIN A, ANAND S, VLAHOS P, CHANDRAJITH R & NANAYAKKARA N 2023a. Describing Natural History and Exploring Risk Factors for Kidney Function Decline in Persons With CKD of Uncertain Etiology in Sri Lanka. Kidney international reports [Online], 8. Available: http://europepmc.org/abstract/MED/37441476 http://www.kireports.org/article/S2468024923012652/pdf 10.1016/j.ekir.2023.04.010 https://europepmc.org/articles/PMC10334401 https://europepmc.org/articles/PMC10334401?pdf=render [Accessed 2023/07//]. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. HEWAVITHARANA P, SCHENSUL S, LEE E, MONTEZ-RATH M, SENARATHNE S, LIU S, HAROLD K, HEWAPATHIRANAGE S, ERANDIKA N, ABEYSUNDARA HTK, YU X, BHALLA V, FIRE A, LEVIN A, ANAND S, VLAHOS P, CHANDRAJITH R & NANAYAKKARA N 2023b. Describing Natural History and Exploring Risk Factors for Kidney Function Decline in Persons With CKD of Uncertain Etiology in Sri Lanka. Kidney Int Rep, 8, 1430–1438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. JANOUDI G, UZUN RADA M, FELL DB, RAY JG, FOSTER AM, GIFFEN R, CLIFFORD T & WALKER MC 2024. Outlier analysis for accelerating clinical discovery: An augmented intelligence framework and a systematic review. PLOS Digit Health, 3, e0000515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. JAYASINGHE S, LIND L, SALIHOVIC S, LARSSON A & LIND PM 2018. High serum levels of p,p’-DDE are associated with an accelerated decline in GFR during 10 years follow-up. Sci Total Environ, 644, 371–374. [DOI] [PubMed] [Google Scholar]
  22. JAYASUMANA C, GUNATILAKE S & SENANAYAKE P 2014. Glyphosate, hard water and nephrotoxic metals: are they the culprits behind the epidemic of chronic kidney disease of unknown etiology in Sri Lanka? Int J Environ Res Public Health, 11, 2125–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. JAYASUMANA C, PARANAGAMA P, AGAMPODI S, WIJEWARDANE C, GUNATILAKE S & SIRIBADDANA S 2015. Drinking well water and occupational exposure to Herbicides is associated with chronic kidney disease, in Padavi-Sripura, Sri Lanka. Environ Health, 14, 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. KEOGH SA, LEIBLER JH, SENNETT DECKER CM, AMADOR VELÁZQUEZ JJ, JARQUIN ER, LOPEZ-PILARTE D, GARCIA-TRABANINO R, DELGADO IS, PETROPOULOS ZE, FRIEDMAN DJ, AMADOR SÁNCHEZ MR, GUEVARA R, MCCLEAN MD, BROOKS DR & SCAMMELL MK 2022. High prevalence of chronic kidney disease of unknown etiology among workers in the Mesoamerican Nephropathy Occupational Study. BMC Nephrol, 23, 238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. KUPFERMAN J, RAMÍREZ-RUBIO O, AMADOR JJ, LÓPEZ-PILARTE D, WILKER EH, LAWS RL, SENNETT C, ROBLES NV, LAU JL, SALINAS AJ, KAUFMAN JS, WEINER DE, SCAMMELL MK, MCCLEAN MD, BROOKS DR & FRIEDMAN DJ 2018. Acute Kidney Injury in Sugarcane Workers at Risk for Mesoamerican Nephropathy. Am J Kidney Dis, 72, 475–482. [DOI] [PubMed] [Google Scholar]
  26. LITTLE R, RUBIN D 2002. Statistical Analysis with Missing Data, John Wiley & Sons, Inc. [Google Scholar]
  27. LIYANAGE DND, DIYABALANAGE S, DUNUWEERA SP, RAJAPAKSE S, RAJAPAKSE RMG & CHANDRAJITH R 2022. Significance of Mg-hardness and fluoride in drinking water on chronic kidney disease of unknown etiology in Monaragala, Sri Lanka. Environ Res, 203, 111779. [DOI] [PubMed] [Google Scholar]
  28. LO RS, CHAN JC, COCKRAM CS & LAI FM 1995. Acute tubular necrosis following endosulphan insecticide poisoning. J Toxicol Clin Toxicol, 33, 67–9. [DOI] [PubMed] [Google Scholar]
  29. MAKITA Y, MATSUURA T, OGATA R, ROMERO Y, OMURA M, TANAKA A, HIRATA M & INOUE N 2003. Systemic effects of orally administered p, p’-DDE on immature male Wistar rats during pubertal period. J Occup Health, 45, 223–7. [DOI] [PubMed] [Google Scholar]
  30. MONTEZ-RATH ME, WINKELMAYER WC & DESAI M 2014. Addressing missing data in clinical studies of kidney diseases. Clin J Am Soc Nephrol, 9, 1328–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. NANAYAKKARA S, SENEVIRATHNA ST, PARAHITIYAWA NB, ABEYSEKERA T, CHANDRAJITH R, RATNATUNGA N, HITOMI T, KOBAYASHI H, HARADA KH & KOIZUMI A 2015. Whole-exome sequencing reveals genetic variants associated with chronic kidney disease characterized by tubulointerstitial damages in North Central Region, Sri Lanka. Environ Health Prev Med, 20, 354–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. NANAYAKKARA S, SENEVIRATHNA STMLD, HARADA KH, CHANDRAJITH R, HITOMI T, ABEYSEKERA T, MUSO E, WATANABE T & KOIZUMI A 2019. Systematic evaluation of exposure to trace elements and minerals in patients with chronic kidney disease of uncertain etiology (CKDu) in Sri Lanka. Journal of Trace Elements in Medicine and Biology, 54, 206–213. [DOI] [PubMed] [Google Scholar]
  33. NANAYAKKARA S, SENEVIRATHNA STMLD, HARADA KH, CHANDRAJITH R, NANAYAKKARA N & KOIZUMI A 2020. The Influence of fluoride on chronic kidney disease of uncertain aetiology (CKDu) in Sri Lanka. Chemosphere, 257, 127186. [DOI] [PubMed] [Google Scholar]
  34. RANASINGHE AV, KUMARA G, KARUNARATHNA RH, DE SILVA AP, SACHINTANI KGD, GUNAWARDENA J, KUMARI S, SARJANA MSF, CHANDRAGUPTHA JS & DE SILVA MVC 2019. The incidence, prevalence and trends of Chronic Kidney Disease and Chronic Kidney Disease of uncertain aetiology (CKDu) in the North Central Province of Sri Lanka: an analysis of 30,566 patients. BMC Nephrol, 20, 338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. RANGO T, JEULAND M, MANTHRITHILAKE H & MCCORNICK P 2015. Nephrotoxic contaminants in drinking water and urine, and chronic kidney disease in rural Sri Lanka. Sci Total Environ, 518–519, 574–85. [DOI] [PubMed] [Google Scholar]
  36. REID M 2022. John Snow Hunts the Blue Death. Distillations Magazine. [Google Scholar]
  37. ROJAS-LIMA E, ORTEGA-ROMERO M, AZTATZI-AGUILAR OG, RUBIO-GUTIÉRREZ JC, NARVÁEZ-MORALES J, ESPARZA-GARCÍA M, MÉNDEZ-HERNÁNDEZ P, MEDEIROS M & BARBIER OC 2025. Vanadium exposure and kidney markers in a pediatric population: a cross-sectional study. Pediatr Nephrol, 40, 1689–1700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. SHAH MD & IQBAL M 2010. Diazinon-induced oxidative stress and renal dysfunction in rats. Food Chem Toxicol, 48, 3345–53. [DOI] [PubMed] [Google Scholar]
  39. SHILLER AM 1997. Dissolved trace elements in the Mississippi River: Seasonal, interannual, and decadal variability. Geochimica et Cosmochimica Acta, 61, 4321–4330. [Google Scholar]
  40. SMPOKOU ET, GONZÁLEZ-QUIROZ M, MARTINS C, ALVITO P, LE BLOND J, GLASER J, ARAGÓN A, WESSELING C, NITSCH D, PEARCE N, NORMAN J, LINDH CH, MORTON J & CAPLIN B 2019. Environmental exposures in young adults with declining kidney function in a population at risk of Mesoamerican nephropathy. Occup Environ Med, 76, 920–926. [DOI] [PubMed] [Google Scholar]
  41. SRIVASTAVA S & FLORA SJS 2020. Fluoride in Drinking Water and Skeletal Fluorosis: a Review of the Global Impact. Curr Environ Health Rep, 7, 140–146. [DOI] [PubMed] [Google Scholar]
  42. STATACORP 2024. Stata 18 Base Reference Manual, College Station, Tx: Stata Press. [Google Scholar]
  43. UCIBIOR A, GOŁĘBIOWSKA D, ADAMCZYK A, NIEDŹWIECKA I & FORNAL E 2014. The renal effects of vanadate exposure: potential biomarkers and oxidative stress as a mechanism of functional renal disorders--preliminary studies. Biomed Res Int, 2014, 740105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. ULRICH JC, HOFFMAN K, GUNASEKARA TDKSC, SANDAMINI PMMA, JACKSON BP, DE SILVA PMCS, JAYASUNDARA N & FERGUSON PL 2023. Glyphosate and Fluoride in High-Hardness Drinking Water Are Positively Associated with Chronic Kidney Disease of Unknown Etiology (CKDu) in Sri Lanka. Environmental Science & Technology Letters, 10, 916–923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. URBAN ARAGON JA, BANDYOPADHYAY E, FERNANDO AS, DE LA FUENTE CASTRO C, WELIKALA AHJ, BIDDANDA A, WITONSKY D, SANDER N, KOTELAWALA JT, PASUPULETI N, STEINRÜCKEN M, ADIKARI G, TENNEKOON K, RAGSDALE AP, TERHORST J, RANASINGHE R, RAI N & RAGHAVAN M. 2025. Population histories of the Indigenous Adivasi and Sinhalese from Sri Lanka using whole genomes. Current Biology, 35, 2554–2566.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. US ENVIRONMENTAL PROTECTION AGENCY 2016. Endosulfan Phase-out. [Google Scholar]
  47. VALCKE M, LEVASSEUR M-E, SOARES DA SILVA A & WESSELING C 2017. Pesticide exposures and chronic kidney disease of unknown etiology: an epidemiologic review. Environmental Health, 16, 49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. VLAHOS P, SCHENSUL SL, ANAND S, SHIPLEY E, DIYABALANAGE S, HU C, HA T, STANIEC A, HAIDER L, SCHENSUL JJ, HEWAVITHARANE P, SILVA T, CHANDRAJITH R & NANAYAKKARA N 2021. Water sources and kidney function: investigating chronic kidney disease of unknown etiology in a prospective study. npj Clean Water, 4, 50. [Google Scholar]
  49. VLAHOS P, SCHENSUL SL, NANAYAKKARA N, CHANDRAJITH R, HAIDER L, ANAND S, SILVA KT & SCHENSUL JJ 2019. Kidney progression project (KiPP): Protocol for a longitudinal cohort study of progression in chronic kidney disease of unknown etiology in Sri Lanka. Glob Public Health, 14, 214–226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. WARNAKULASURIYA KA, BALASURIYA S, PERERA PA & PEIRIS LC 1992. Determining optimal levels of fluoride in drinking water for hot, dry climates--a case study in Sri Lanka. Community Dent Oral Epidemiol, 20, 364–7. [DOI] [PubMed] [Google Scholar]
  51. WHITFORD G, PASHLEY D & STRINGER G 1976. Fluoride renal clearance: a pH-dependent event. American Journal of Physiology-Legacy Content, 230, 527–532. [DOI] [PubMed] [Google Scholar]
  52. WIMALAWANSA SJ 2014. Escalating chronic kidney diseases of multi-factorial origin in Sri Lanka: causes, solutions, and recommendations. Environ Health Prev Med, 19, 375–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. WORLD HEALTH ORGANIZATION 2019. The WHO Recommended Classification of Pesticides by Hazard and guidelines to classification, 2019 edition. [Google Scholar]

Associated Data

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

Supplementary Materials

Supplemental material
NIHMS2171839-supplement-Supplemental_material.docx (47.8KB, docx)

Data Availability Statement

Data may be made available upon reasonable request following the completion of data sharing approvals from all three collaborating institutions: Stanford University, the University of Connecticut, and the National Teaching Hospital in Kandy, Sri Lanka.

RESOURCES