CKDu-AGRI Study—a population-based cross-sectional study of chronic kidney disease and chronic kidney disease of unknown aetiology among agricultural workers in Tamil Nadu, India

Summary

Background

Chronic kidney disease of unknown cause (CKDu) has been reported in “hotspots” around the world and linked to occupational heat stress, exposure to agrochemicals and environmental toxins. This was a systematic, statewide, population-based, observational study of agricultural workers in Tamil Nadu, to estimate the prevalence of chronic kidney disease (CKD) and CKDu.

Methods

The state of Tamil Nadu was divided into five agroclimatic zones and stratified multi-stage cluster sampling was performed to select a representative study population. All participants underwent clinical evaluation and laboratory testing (Phase 1). Repeat sampling was performed after 3 months for all individuals with an eGFR <60 mL/min per 1.73 m² during initial sampling (Phase 2).

Findings

Among 3350 participants who were screened, the number of participants with eGFR ≤60 mL/min per 1.73 m² at the end of Phase 1 and Phase 2 were 584 and 178, respectively. The overall prevalence of CKD was therefore 5.31% (95% CI 4.58–6.13), with the prevalence of CKDu being 2.66% (95% CI 2.14–3.26). The factors that were significantly associated with CKD were increased age, diabetes, hypertension, absence of formal education, anaemia, smokeless tobacco use, and weekly hours of outdoor work. Notably, 406 participants with an eGFR ≤60 mL/min per 1.73 m² during Phase 1 had a normal eGFR when re-measured during Phase 2. These episodes of transient subclinical AKI could potentially contribute to CKD.

Interpretation

The prevalence of CKD among agricultural workers in Tamil Nadu is 5.31%, with about half of these due to CKDu. The effect of transient subclinical AKI on CKD progression needs further study.

Funding

Tamil Nadu Health System Reform Program, Department of Health and Family Welfare, Government of Tamil Nadu.

Research in context.

Evidence before this study

The studies on chronic kidney disease (CKD) or CKD of unknown etiology done in agriculture communities are of small cohort studies or confined to geographic hot spots. Studies restricted to certain areas or population have limitation in the estimate of true prevalence and have inherent selection bias. Further, such studies will have limitations in understanding the phenotypes between the affected individuals and those remain unaffected. On the other hand, a population-based study across wider geographical regions will overcome the bias and would provide more information on the susceptibility factors at level of individual subjects towards CKDu. This is the first study done in different agroclimatic zones of Tamilnadu with robust sampling technique and central reference lab for creatinine measurement by standardised method. Further we repeated the creatinine measurement after three months to confirm the diagnosis of CKD as per KDIGO guidelines.

Added value of this study

In this study, we found CKD prevalence of 5.31% among 3350 agriculture workers, of which CKD of unknown cause contributed to half (2.66%) of the CKD burden. Higher age, male sex, lack of formal education, diabetes, hypertension, history of high cholesterol, anaemia, exposure to smokeless tobacco and long hours of outdoor work were found to be associated with CKD. Except nocturia, the other symptoms related to CKD like edema were present only in less than 2% of participants, thereby delays the individual in seeking medical advice. This points to a need for proactive screening strategy at fixed intervals to identify the CKDu at an early stage among agriculture workers. We observed a phenomenon of asymptomatic transient AKI in 17.4% of the study population, nearly three times the number of CKD. As there is enough literature evidence for AKI to CKD progression these subjects with subclinical AKI have high risk of future CKD. Though it is a questionnaire-based method, we quantified the weekly work hours and found a significant association between number of work hours between those with near normal renal function and CKD and CKDu.

Implications of all the available evidence

Identification of individual risk factors for such subclinical AKI phenomenon and exposure factors like heat, water intake and agrochemicals are needed through future longitudinal studies. Future workplace intervention studies targeting implementation of water, rest and shade strategies and regulated hours of hot environmental exposure are needed to revise and introduce new occupational health guidelines to prevent CKD and CKDu among agriculture workers.

Introduction

Chronic kidney disease (CKD) is a major contributor to mortality and morbidity from non-communicable diseases (NCDs). The Global Burden of Diseases, Injuries, and Risk Factors Study reported that the global all-age prevalence and mortality from CKD increased from 29.3 to 41.5%, between 1990 and 2017.¹ The causes of CKD should be urgently addressed if we are to achieve the United Nations’ 2030 Agenda for Sustainable Development goal of reducing the premature mortality from NCDs by a third.

Global scenario in CKDu

The first cluster of CKDu was identified among agricultural communities of El Salvador,² the aetiology was not attributed to diabetes, hypertension or proteinuric primary glomerular diseases. Subsequently, multiple cross-sectional studies done in various regions of Mexico, Nicaragua and Panama revealed similar clinical phenotype of CKD among sugarcane, corn and banana field workers as well as among construction workers and got the geographical tag of Mesoamerican nephropathy.³ Subsequent evidence from Taiwan⁴ and North Central Province of Sri Lanka⁵ showed similar pattern of CKD among agricultural communities, thus the term Mesoamerican nephropathy was renamed as CKD of unknown aetiology. Despite two decades of epidemiological, cohort and case–control studies from Latin America to Asia, except for the clinicopathological phenotype showing some uniformity, aetiopathogenesis of the disease remains unknown. The case definition of CKDu proposed by the Sri Lankan Society of Nephrology in 2018⁶ has been widely accepted as the standard definition for possible, probable and definitive diagnosis of CKDu.

Indian scenario

The Indian CKD Registry is a multicentre, prospective cohort study of patients with mild-to-moderate CKD from 11 large centres. Although diabetes remained the most common cause of CKD (24.9%), this was followed closely by CKD of unknown cause (CKDu) (19.5%).⁷ In a secondary data analysis of three population-based studies conducted by O'Callaghan-Gordo and colleagues, it was found that CKDu was highly prevalent among rural populations of southern India.⁸ John and colleagues had reported a lack of information on the CKDu phenotype in India and had recommended community-based research that should study the entire CKD population, rather than restricting research only to cases with presumed CKDu based on predefined criteria.⁹

Tamil Nadu, the southernmost state in India, has reported clusters of CKDu.⁹ A study from a referral centre in Pondicherry on a hospital-based registry, found that 52% of the patients in their CKD registry had CKDu. The diagnosis of CKDu was made, as a diagnosis of exclusion, in the absence of any of these potentially identifiable causes of CKD and when the blood pressure <160/100 mm Hg or if the patient required only two or fewer antihypertensive drugs. A geographical mapping of these cases showed that more than half of them were from the districts of Villupuram and Cuddalore in the state of Tamil Nadu. Nearly 53.6% of the affected individuals reported farming as their occupation.¹⁰ Similarly, in the Uddanam region of Andhra Pradesh a multi-stage cluster random sampling from 67 villages involving 2419 participants had an overall CKD prevalence of 10.24%. Older age, male sex, tobacco use, hypertension, and family history of CKD were independently associated with CKD. The study also found a high prevalence of hypertension (42%) compared to national surveys.¹¹

Researches aimed at finding the aetiology for this enigmatic disease has documented evidence on agrochemicals like glyphosate contamination in soil and water resources and cadmium exposure in Sri Lanka and exposure to silica in Uddanam region. To date there is no definitive evidence for links between agrochemicals or heavy metals in cohort studies. One common denominator in high-prevalence regions in Latin America and Asia is exposure to heat stress with the hypothesis that high ambient temperature along with high-intensity manual work and limited access to drink water make the kidney vulnerable to repetitive insults of transient acute kidney injury (AKI).^12,13 However, most of the evidence accrues from observational studies in tertiary referral hospitals that might have been subject to referral bias and therefore not representative of the general population or small cohort studies in the working environment or community-based study with random sampling and single measurement of serum creatinine. Therefore, a community-based study with rigorous sampling methods and a repeat follow-up measurement of creatinine to confirm the diagnosis of CKD according to KDIGO definition is required to estimate the true burden of CKD and CKDu and to understand its phenotype and associated factors among agricultural workers.

This study was conducted among agricultural workers from different agroclimatic zones across the entire state of Tamil Nadu. The aim was to estimate the prevalence of, CKD and CKDu and to find associations of sociodemographic factors, comorbid conditions and outdoor work exposure to CKD and CKDu among agricultural workers.

Methods

This was a systematic, statewide, population-based, cross-sectional observational study conducted among 3350 agricultural workers across different agroclimatic zones of Tamil Nadu between August and December 2023. The state of Tamil Nadu was divided into seven agroclimatic zones based on the rainfall pattern, altitude, and irrigation sources (Fig. 1a). However, for operational convenience, we combined the high-altitude zone with the north western zone and the high rainfall zone with the southern zone and created five zones. Adults aged ≥18 years, whose predominant occupation was farming-related manual labour, and who were actively involved in field work for more than a year were included. There was no specific exclusion except for those who did not give consent to participate in the study.

Fig. 1.

(a) Seven agroclimatic zones of Tamil Nadu state (b) Prevalence of CKD and CKDu in five different agroclimatic zones.

Operational definition

CKD was defined as eGFR ≤60 mL/min per 1.73 m² by the CKD-EPI (2021) creatinine equation and persisting for more than 3 months.

CKDu was defined as per the Sri Lankan Society of Nephrology guideline.⁶

The definition of AKI in the clinical context is defined by a rise of 0.3 mg/dL from baseline creatinine according to KDIGO (Kidney Disease: Improving Global Outcomes) staging in the AKI guidelines. However, because we did not know the baseline creatinine for the subjects we defined transient subclinical AKI as those with eGFR <60 mL/min per 1.73 m² in the first sample but subsequently normalised (ie, eGFR >60 mL/min per 1.73 m²) in the second sample taken after an interval of three months.

Newly detected diabetes was defined as random blood sugar ≥200 mg/dL, and hypertension as blood pressure ≥140/90 mm Hg.

Study procedure

Phase 1

A stratified multi-stage cluster sampling technique was used, as described in Fig. 2 and Table S2. All villages in each agroclimatic zone were serially listed, and 25 villages (‘clusters’) were randomly selected by a computer-generated random number table. The selected villages were visited by the survey team on two separate days. On the first day, the purpose of the survey was explained to the local leaders, and the households with eligible individuals were enumerated. From this list, 30 households were chosen by simple random sampling. If there was more than one eligible member engaged in farming in the chosen household, the KISH method (Fig. S1) was used to select the individual to be included in the study. If the selected individual in the household was unavailable on the day of the visit, the survey team rescheduled another visit at a later date. Importantly, other adult individuals living in the same household, even if eligible, could not be included.

Fig. 2.

Technique of multi-stage cluster sampling.

On the second day, the survey teams visited each individual household for data collection. A total of ten survey teams were engaged for the study, with each team covering 12–13 villages. The survey team comprised one field investigator and one phlebotomist. Informed consent was obtained from each participant. A prespecified semistructured questionnaire was administered by the field investigator in Tamil. This questionnaire included information regarding demography, occupation, family history, comorbidities and other relevant risk factors. The questionnaire was administered by the trained health care provider using EpiCollect5 app. It took 20–30 min on average to complete the questionnaire for each individual.

Anthropometric measurements—height, weight, waist circumference, and blood pressure–were measured by the field staff ensuring adequate privacy. Blood pressure was measured by an automated BP apparatus (Omron, a validated instrument model type according to the local availability). The data collection was done using the EpiCollect5 app. The participants were again asked to provide informed written consent for biological samples, which included urine and blood samples. A urine sample was collected for testing urine albumin and RBCs using on-the-spot dipstick strips. 5 mL of blood was collected by the phlebotomist under aseptic precautions. The collected blood was aliquoted equally into an EDTA tube and gel-barrier tube. Estimation of haemoglobin was done by the cyanmethemoglobin method and random blood sugar by the glucose oxidase-peroxidase method in the semiautomated analyser in the respective primary health centres (PHC) and the blood collected in the gel-barrier tubes was centrifuged in the PHC lab, and the separated serum from all the samples was transported to the State Public Health Laboratory in Chennai in ice-pack-lined thermally stable containers within 24 h. In the central lab, serum creatinine was measured by the Jaffe kinetic method using isotope-dilution mass spectrometry IDMS standards.14, 15, 16

Quality check and data management

The project was built as a private project and only people who were assigned specific roles in the project had access to the form. There were specific assigned roles and those with the data-collector role were not provided access to entries made by others. The questionnaire was developed in a workshop format with experts from Community Medicine and Nephrology. The tool was piloted before use. The teams were instructed to upload their data on a daily basis, incoming data were checked for consistency and various data validation checks were used to ensure data accuracy. The following quality assurance measures were implemented to ensure the quality of field data collection. All team members were trained rigorously on interview techniques and the use of the equipment. All interview responses were recorded electronically and there was no need for manual data entry. The questionnaire was precoded with specific check codes and skip patterns to minimise entry errors using EpiCollect5. Each individual was assigned a cluster ID (three-digit: 001–125) and an individual ID (two-digit: 01–30) generated in the app and the same was manually entered in their biological samples.

Phase 2

This included only the participants from Phase 1 who were identified to have an eGFR ≤60 mL/min per 1.73 m² by the CKD-EPI (2021) creatinine equation and/or a dipstick urine protein ≥1+. Survey teams undertook another visit to the respective households, 3 months after Phase 1. Repeat blood and urine samples were collected and analysed as specified earlier, in order to confirm the persistence of the laboratory abnormalities that were detected in Phase 1.

Additionally, participants from Phase 1 whose blood samples were unintentionally lysed and unsuitable for analysis were also resampled in Phase 2 (Table S2).

The Universal Thermal Climate Index (UTCI) heat-stress model

UTCI is an equivalent temperature that characterizes the physiological response of the human body to a meteorological input.¹⁷ UTCI is defined using the temperature, humidity, wind, radiation, metabolic rate and clothing using a 187-node human bioclimatic model. This ensures a more accurate representation of the effect of heat stress on the human body. To model the UTCI for the agroclimatic zones of Tamil Nadu, hourly UTCI data were sourced from the Copernicus platform, 0.25° × 0.25° (27.5 × 27.5 km) grid resolution from the ERA5 HEAT Thermal Indices gridded dataset.¹⁸ The corresponding NetCDF-4 (Network Common Data Form) files for the year 2023 were obtained for the geographical boundaries of Tamil Nadu. The files were then processed using Python libraries to generate the following plots daytime average UTCI for the three months-July, August and December 2023 and days of the month when UTCI is > 38 °C for >4 h.

Ethics committee approval

The survey protocol was approved by the Scientific and Ethics Committee of the Madras Medical College Institutional Review Board, approval number 0101/2023 dated 05/01/23 and Institutional ethical committee of Directorate of Public Health and Preventive medicine, approval number DPHPM/IEC/093/2023 dated 24/02/2023. Informed consent in Tamil was obtained from every survey participant after explaining the details of the project and requirement of biological samples in their native language (Tamil). Participants were given the results of physical measurements and laboratory results. If abnormal results were encountered, the patients were referred to the nearest Government Hospital for treatment and follow-up care.

Sample size calculation

The sample size was calculated assuming an overall pooled prevalence of CKD in India of 10.2% as per reported literature.¹⁹ Assuming a level of confidence (for α = 0.05 and 95% confidence level), a margin of error of 20% of prevalence, and after correcting for the multi-stage sampling method with a design effect of 1.5 and a non-response rate of 20%, the required sample size was estimated to be 3666.

Statistical analysis

Statistical analyses were performed using Stata (version 17), incorporating survey design adjustments to account for the complex sampling structure. Although probability proportional to size (PPS) sampling was initially planned, village-level population data were not available to construct sampling weights. Therefore, formal PPS weighting was not implemented. Nevertheless, survey-adjusted analyses were carried out using Stata's svyset and associated commands, specifying the village as the primary sampling unit (PSU). This approach accounts for clustering effects and provides robust standard errors and valid statistical inference. Survey-adjusted methods were used to estimate means, proportions, odds ratios, and corresponding 95% confidence intervals (CIs).

Descriptive statistics were used to summarise the baseline characteristics of the study population. Continuous variables were presented as means with standard deviations (SD) for normally distributed data. Categorical variables were expressed as frequencies and percentages with 95% CIs.

For comparison between groups, survey-weighted chi-squared tests were used to assess associations between categorical variables. Where cell counts were small, Fisher's exact test was applied as appropriate. For continuous variables, comparisons were made using survey-adjusted t-tests for normally distributed data to determine statistical significance.

To identify factors associated with chronic kidney disease (CKD) and CKD of unknown aetiology (CKDu), survey-weighted logistic regression models were applied. Univariable logistic regression was used to estimate crude odds ratios (ORs) with 95% confidence intervals (CIs). For the multivariable logistic regression, the variables with p < 0.20 in univariable analyses were included in the models to control for potential confounders. The results were reported as adjusted odds ratios (aORs) with 95% CIs. Age and weekly outdoor work hours were categorised using cut-offs based on exploratory data analysis, optimising discriminative performance for CKD and CKDu versus normal renal function based on the Youden index.

Model diagnostics were performed to assess multicollinearity using variance inflation factors (VIFs), and model fit was evaluated through appropriate goodness-of-fit tests. All statistical tests were two-sided, and p < 0.05 was considered statistically significant. Participants with missing data for primary outcomes or key covariates were excluded from relevant analyses. As the overall proportion of missing data was low (<2%), the risk of bias was minimal. Non-response weighting or imputation were not applied.

Role of the funding source

The funding agent had no role in study design, data collection, data analysis, interpretation and writing of the report.

Results

CKD prevalence and clinical characteristics

In Phase 1 of the study, conducted in September 2023, a total of 3750 participants were approached, of whom 3350 were eventually included. The distribution of study participants across different agroclimatic zones and other clinical and demographic characteristics are described in Fig. 1b and Table 1. An eGFR ≤60 mL/min per 1.73 m² was noted in 17.43% (584/3350) of the study participants.

Table 1.

Baseline demographic, clinical, and occupational characteristics of adult agriculture workers in Tamil Nadu, India (N = 3350).

	Number of participants (n)	Percentage (%)
Age in years (mean ± SD)	50.9 ± 14.3
Sex
Women	1749/3350	52.2
Men	1601/3350	47.7
Education
Illiterate	1096/3347	32.7
Able to read/write (but no formal education)	405/3347	12.1
Primary school	882/3347	26.4
High school	515/3347	15.4
Higher secondary school	292/3347	8.7
Diploma/Graduate	157/3347	4.7
Socio-economic status (BG Prasad Scale 2022)
Lower	12/3346	0.4
Lower middle	158/3346	4.7
Middle	169/3346	5.0
Upper middle	1045/3346	31.2
Upper	1962/3346	58.6
Source of water
Bore well	564/3347	16.8
Municipal/panchayat water	2700/3347	80.7
Can water	70/3347	2.1
Others	13/3347	0.3
Median week-hours of outdoor work (IQR)	35 (24–48)
Individuals who worked outdoors in the past	204/3347	6.1
Individuals who worked overseas in the past	58/3347	1.7
Previously known diabetes mellitus	276/3347	8.2
Newly detected diabetes mellitus	175/3347	5.2
Years of treatment for diabetes mellitus
≤5 years	163/275	59.3
6–10 years	67/275	24.4
>10 years	45/275	16.5
Previously known hypertension	326/3347	9.7
Newly detected hypertension	681/3347	20.3
Years of treatment for hypertension
≤5 years	202/324	62.3
6–10 years	81/324	25.0
>10 years	41/324	12.7
Prior heart disease	33/3347	1.0
Prior cerebrovascular accident	13/3347	0.4
Dyslipidemia	36/3347	1.1
History of snake bite	38/3347	1.1
History of Covid-19	301/3347	9.0
History of smoking
Never	3092/3347	92.4
Past	103/3347	3.1
Present	152/3347	4.5
History of smokeless tobacco use
Never	3184/3347	95.1
Past	50/3347	1.5
Present	113/3347	3.4
History of alcohol consumption
Never	2980/3347	89.0
Past	126/3347	3.8
Present	241/3347	7.2
Type of diet
Vegetarian diet	519/3347	15.5
Non-vegetarian diet ≤2 days	2250/3347	67.2
Non-vegetarian diet 3–5 days	535/3347	16.0
Non-vegetarian diet >5 days	43/3347	1.3
Self-reported consumption of traditional medicines	14/3347	0.4
Self-reported consumption of over-the-counter analgesics	120/3347	3.6
BMI (Asian classification)
<18.5
Underweight	377/3334	11.3
18.5–22.9
Normal	1238/3334	37.1
23–24.9
Overweight	617/3334	18.5
25–29.9
Pre-obese	836/3334	25.1
≥30
Obese	266/3334	8.0
Waist circumference (women)
Normal (≤80.9 cm)	993/1748	56.8
High (81.0–89.9 cm)	346/1748	19.8
Very high (≥90 cm)	409/1748	23.4
Waist circumference (men)
Normal (≤94.9 cm)	1321/1592	83.0
High (95.0–101.9 cm)	192/1592	12.1
Very high (≥102 cm)	79/1592	4.9
Self-reported symptoms of kidney disease
Pedal oedema	43/3347	1.3
Facial puffiness	12/3347	0.4
Nocturia (>2 times per night)	433/3347	12.9
Recurrent UTI (>2 episodes per year)	13/3347	0.4
Renal calculi in the last 5 years	25/3347	0.7
Patients on treatment for ESKD
Maintenance hemodialysis	12/3347	0.4
Kidney transplantation	9/3347	0.3

Phase 2 of the study was conducted in December 2023, and 642 individuals had a re-measurement of serum creatinine. Of these, only 178 participants had a persistently low eGFR ≤60 mL/min per 1.73 m², fulfilling the criteria for CKD. Thus, the overall prevalence of CKD and CKDu among the agricultural population was 5.31% (95% CI 4.58–6.13) and 2.66% (95% CI 2.14–3.26) respectively, with the highest prevalence of CKD (7.7%) in the north-east zone and lowest in the north-west zone (2.16%) Fig. 1b. The participant flow diagram is presented in Fig. 3.

Fig. 3.

Participant flow diagram.

The majority of participants were asymptomatic. Notably, however, increased nocturnal frequency of micturition (more than two times a night) was reported by 12.94% (n = 433) of the participants. In total, 34 participants with CKD had ≥1+ proteinuria, of whom, 11 had CKDu.

Factors associated with CKD: univariable analysis

Univariable analysis of risk factors for CKD is described in Table 2. Age was significantly associated with CKD [OR 1.03 (95% CI, 1.02–1.05; p < 0.001)]. A significant association among the social factors for CKD was absence of formal education, in whom the prevalence was 7.9% [OR 1.08 (95% CI, 0.62–1.89; p = 0.76)] compared to 2.5% [OR 0.33 (95% CI, 0.11–0.96; p = 0.043)] among the graduates. The most common traditional risk factors were diabetes [OR 2.42 (95% CI, 1.56–3.75; p < 0.001)] and hypertension [OR 3.17 (95% CI, 2.18–4.60; p < 0.001)], with the CKD prevalence of 10.9% and 12.1% among them respectively. The additional non-traditional risk factors were the number of week-hours of outdoor work [OR 1.00 (95% CI, 1.00–1.01; p = 0.039)] and exposure to smokeless tobacco [OR 2.36 (95% CI, 1.33–4.18; p = 0.004)]

Table 2.

Factors associated with chronic kidney disease (CKD) among the agriculture workers.

	Participants without kidney disease (n = 3172)		Participants with CKD (n = 178)		OR (95% CI)	P-value
	n	Percentage (95% CI)	n	Percentage (95% CI)
Age in years (Mean ± SD)	56.8 ± 13.5		62.8 ± 10.6		1.03 (1.02, 1.05)	<0.001
Sex
Women	1664	95.1 (93.7, 96.3)	85	4.9 (3.7, 6.3)	1	0.256
Men	1508	94.2 (92.5, 95.5)	93	5.8 (4.4, 7.5)	1.20 (0.87, 1.67)
Education
Illiterate	1016	92.7 (90.3, 94.5)	80	7.3 (5.5, 9.7)	1
Able to read/write (but no formal education)	373	92.1 (87.6, 95.0)	32	7.9 (4.9, 12.4)	1.08 (0.62, 1.89)	0.76
Primary school	844	95.7 (94.1, 96.9)	38	4.3 (3.1, 5.9)	0.57 (0.39, 0.83)	0.004
High school	285	97.6 (94.0, 99.1)	7	2.4 (0.9, 6.0)	0.31 (0.11, 0.83)	0.021
Higher secondary school	499	96.9 (94.9, 98.1)	16	3.1 (1.8, 5.1)	0.40 (0.22, 0.72)	<0.001
Diploma/Graduate	153	97.4 (93.3, 99.1)	4	2.5 (0.9, 6.7)	0.33 (0.11, 0.96)	0.043
Source of water
Borewell	544	96.4 (93.5, 98.1)	20	3.5 (1.9, 6.5)	1
Municipal/panchayat water	2547	94.3 (92.9, 95.5)	153	5.7 (4.5, 7.1)	1.63 (0.84, 3.15)	0.142
Can water	67	95.7 (85.4, 98.8)	3	4.3 (1.1, 14.6)	1.21 (0.27, 5.40)	0.794
Others	12	92.3 (94.7, 96.3)	1	7.7 (3.7, 15.3)	2.26 (0.87, 5.85)	0.09
Mean week-hours of outdoor work (Mean ± SD)	33.4 ± 17.6		36.2 ± 18.1		1.00 (1.00, 1.01)	0.039
Individuals who worked outdoors in the past
No	2984	94.9 (93.7, 96.0)	159	5.0 (4.0, 6.3)	1
Yes	186	91.2 (82.8, 95.7)	18	8.8 (4.3, 17.2)	1.81 (0.82, 4.01)	0.139
Individuals who worked overseas in the past
No	3115	94.7 (93.4, 95.7)	174	5.3 (4.3, 6.5)	1
Yes	55	94.8 (81.2, 98.7)	3	5.2 (1.3, 18.8)	0.97 (0.23, 2.07)	0.974
Previously known diabetes mellitus
No	2924	95.2 (94.1, 96.1)	147	4.8 (3.8, 5.9)	1
Yes	246	89.1 (83.7, 92.9)	30	10.9 (7.1, 16.3)	2.42 (1.56, 3.75)	<0.001
Years of treatment for diabetes
≤5 years	150	92.0 (84.9, 95.9)	13	8.0 (4.0, 15.1)	1
6–10 years	57	85.1 (74.7, 91.6)	10	14.9 (8.3, 25.2)	2.02 (0.87, 4.69)	0.1
>10 years	38	84.4 (71.5, 92.2)	7	15.6 (7.8, 28.5)	2.12 (0.75, 5.94)	0.149
Previously known hypertension
No	2757	95.8 (94.8, 96.7)	120	4.2 (3.3, 5.2)	1
Yes	413	87.9 (83.3, 91.3)	57	12.1 (8.7, 16.7)	3.17 (2.18, 4.60)	<0.001
Years of treatment for hypertension
≤5 years	178	88.1 (82, 92.3)	24	11.9 (7.6, 18)	1
6–10 years	68	83.9 (75.4, 89.9)	13	16.0 (10.1, 24.6)	1.41 (0.74, 2.68)	0.28
>10 years	31	75.6 (65.8, 83.3)	10	24.4 (16.8, 34.2)	2.39 (1.24, 4.69)	0.012
Prior heart disease
No	3140	94.7 (93.5, 95.8)	174	5.2 (4.2, 6.5)	1
Yes	30	90.9 (77.0, 96.8)	3	9.1 (3.2, 23.0)	1.80 (0.57, 5.67)	0.31
Prior cerebrovascular accident
No	3159	94.7 (93.5, 95.8)	175	5.2 (4.2, 6.5)	1
Yes	11	84.6 (56.9, 95.8)	2	15.4 (4.2, 43.1)	3.28 (0.79, 13.55)	0.1
Dyslipidemia
No	3139	94.8 (93.5, 95.8)	172	5.2 (4.2, 6.4)	1
Yes	31	86.1 (71.2, 94.0)	5	13.9 (6.0, 28.8)	2.94 (1.17, 7.35)	0.021
History of snake bite
No	3135	94.7 (93.5, 95.8)	174	5.2 (4.2, 6.5)	1
Yes	35	92.1 (73.9, 98.0)	3	7.9 (2.0, 26.1)	1.54 (0.38, 6.14)	0.535
History of Covid-19 infection
No	2891	94.9 (93.6, 96.0)	155	5.1 (4.0, 6.4)	1
Yes	279	92.7 (88.4, 95.5)	22	7.3 (4.5, 11.6)	1.47 (0.84, 2.56)	0.171
History of smoking
Never	2933	94.9 (93.6, 95.9)	159	5.1 (4.1, 6.4)	1
Past	142	93.4 (87.2, 96.7)	10	6.6 (3.2, 12.8)	1.29 (0.64, 2.59)	0.456
Present	95	92.2 (84.1, 96.4)	8	7.8 (3.6, 15.9)	1.55 (0.66, 3.60)	0.302
History of smokeless tobacco use
No	3025	95.0 (93.8, 96.0)	159	5.0 (4.0, 6.1)	1
Yes	145	89.0 (81.4, 93.7)	18	11.0 (6.3, 18.5)	2.36 (1.33, 4.18)	0.004
History of alcohol consumption
No	2829	94.9 (93.7, 95.9)	151	5.1 (4.1, 6.3)	1
Yes	341	92.9 (88.7, 95.6)	26	7.1 (4.4, 11.3)	1.42 (0.86, 2.35)	0.161
Type of diet
Vegetarian	486	93.6 (90.7, 95.7)	33	6.3 (4.3, 9.3)	1
Non-vegetarian	2684	94.9 (93.6, 96.0)	144	5.1 (4.0, 6.4)	0.79 (0.50, 1.22)	0.289
Self-reported consumption of traditional medicines
No	3158	94.7 (93.5, 95.8)	175	5.2 (4.2, 6.5)	1
Yes	12	85.7 (55.1, 96.7)	2	14.3 (3.3, 44.9)	3.00 (0.65, 13.8)	0.155
Self-reported consumption of over-the-counter analgesics
No	3058	94.8 (93.5, 95.8)	169	5.2 (4.2, 6.5)	1
Yes	112	93.3 (87.1, 96.7)	8	6.7 (3.3, 12.9)	1.29 (0.61, 2.71)	0.494
BMI (Asian classification)
Underweight	337	94.4 (91.2, 96.5)	20	5.6 (3.5, 8.8)	0.93 (0.56, 1.55)	0.805
Normal weight	1756	94.0 (92.4, 95.4)	111	5.9 (4.6, 7.6)	1
Over weight	801	95.8 (95.0, 97.1)	35	4.2 (2.9, 6.0)	0.69 (0.46, 1.01)	0.061
Obese	256	96.2 (92.8, 98.1)	10	3.7 (1.9, 7.2)	0.61 (0.30, 1.26)	0.185
Waist circumference
Normal	876	94.4 (92.3, 95.9)	52	5.6 (4.0, 7.7)	1
High	960	93.6 (91.1, 95.4)	66	6.4 (4.6, 8.9)	1.15 (0.77, 1.74)	0.476
Prior episodes of urinary tract infections
≤2 episodes	31	96.9 (78.3, 99.6)	1	3.1 (0.4, 21.7)	1
3–4 episodes	8	80 (35.3, 96.7)	2	20 (3.3, 64.7)	7.75 (0.40, 148.60)	0.166
>4 episodes	3	100	0	0	1
Renal calculi in the last 5 years
No	3150	94.8 (93.6, 95.8)	172	5.2 (4.1, 6.4)	1
Yes	20	80 (61.4, 90.9)	5	20 (9.0, 38.6)	4.57 (1.80, 11.59)	0.002
Renal transplantation
No	3162	94.7 (93.5, 95.8)	176	5.3 (4.2, 6.5)	1
Yes	8	88.9 (67.2, 96.9)	1	11.1 (3.1, 32.8)	2.24 (0.57, 8.78)	0.243
Hemoglobin
>11 g/dL	2322	96.1 (95.0, 96.9)	96	3.9 (3.1, 5.0)	1
<11 g/dL	786	90.8 (87.8, 93.1)	80	9.2 (6.9, 12.2)	2.49 (1.78,3.49)	<0.001

Statistical test used: Logistic regression with significance at the level of 0.05.

Multivariable analysis for CKD

Multivariate analysis of risk factors for CKD is given in Fig. 4. Age, male sex, hypertension, history of renal calculus, anaemia, exposure to smokeless tobacco and long hours of outdoor work were found to have a significant association with CKD.

Fig. 4.

Forest plot of multivariate logistic regression analysis for CKD after adjusting for confounders.

Furthermore, the prevalence of CKD increases with age after 53.5 years [aOR 6.29 (95% CI, 4.15–9.54; p < 0.001)] both in men and women. Outdoor work of more than 24.5 h a week increases CKD risk [aOR 1.78 (95% CI, 1.17–2.72; p = 0.024), as calculated by the Youden index]. The source of potable water, history of snake bite, history of covid infection and alcohol intake had no association with CKD prevalence.

Analysis for CKDu

Overall, 89 (2.66%) participants were classified as CKDu. Higher age, male sex, lack of education, anaemia and long hours of outdoor work were found to be associated with CKDu (Table S4 and Fig. 5.)

Fig. 5.

Forest plot of multivariate logistic regression analysis for CKDu after adjusting for confounders.

Transient subclinical AKI

At the end of Phase 1, a reduced eGFR <60 mL/min per 1.73 m² was noted in 584 participants. In Phase 2, when the serum creatinine was repeated after 3 months, a reduced eGFR <60 mL/min per 1.73 m² was noted only in 178 participants. The remaining 406 participants from Phase 1 were found to have a normalisation of creatinine in Phase 2. As all the samples were measured in a single centre with IDMS standardisation, it is unlikely to have been a measurement error. The only plausible explanation is that these patients had undergone a phenomenon of transient subclinical AKI. As the baseline serum creatinine for the subjects was unknown, we defined transient AKI as eGFR <60 mL/min per 1.73 m² in Phase 1. However, we considered the serum creatinine value of those who recovered from AKI in Phase 2 as baseline, and we proceeded to grade AKI according to KDIGO guidelines. Of those with transient AKI, 374 (92.1%) were in stage 1 AKI, 23 (5.7%) in stage 2 and 9 (2.2%) in stage 3.

A comparison between the patients with transient subclinical AKI and patients with persistent CKD is presented in Table 3. It was found that male sex, diabetes, hypertension, exposure to smokeless tobacco, anaemia and higher mean age were significantly associated with progression to CKD. On comparing those with normal renal function and those with transient AKI, the absence of formal education or only primary-level education, higher mean age and hypertension were found to be associated with subclinical AKI (Table S3).

Table 3.

Comparison of baseline characteristics and risk factors among participants with transient AKI and persistent CKD.

	Participants with transient AKI (n = 406)		Participants with persistent CKD (n = 178)		OR (95% CI)	P-value
	n	Percentage (95% CI)	n	Percentage (95% CI)
Age in years (Mean ± SD)	57.7 ± 13.3		62.8 ± 10.7		1.03 (1.01, 1.04)	<0.001
Sex
Women	231	73.1 (66.5, 78.8)	85	26.9 (21.2, 33.5)	1
Men	175	65.3 (57.2, 72.6)	93	34.7 (27.4, 42.8)	1.44 (0.99, 2.09)	0.053
Education
Illiterate	135	62.8 (53.4, 71.3)	80	37.2 (28.7, 46.6)	1
Able to read/write (but no formal education)	81	71.7 (59.2, 81.5)	32	28.3 (18.5,40.8)	0.66 (0.34, 1.29)	0.229
Primary school	115	75.2 (66.7, 82.0)	38	24.8 (18.0, 33.2)	0.55 (0.33, 0.92)	0.024
High school	22	75.9 (49.7, 90.9)	7	24.1 (9.1, 50.3)	0.53 (0.16, 1.76)	0.303
Higher secondary school	47	74.6 (62.7, 83.7)	16	25.4 (16.3, 37.3)	0.57 (0.29, 1.10)	0.095
Diploma/Graduate	6	60 (27.6, 85.5)	4	40 (14.5, 72.4)	1.12 (0.27, 4.68)	0.87
Mean week-hours of outdoor work (Mean ± SD)	34.2 ± 17.4		36.2 ± 18.1		1.00 (0.99, 1.01)	0.328
Individuals who worked outdoors in the past
No	388	70.9 (64.9, 76.3)	159	29.1 (23.7, 35.1)	1
Yes	18	50 (28.4, 71.6)	18	50 (28.4, 71.6)	2.44 (0.94, 6.31)	0.065
Individuals who worked overseas in the past
No	401	69.7 (63.8, 75.1)	174	30.3 (24.9, 36.2)	1
Yes	5	62.5 (25.1, 89.2)	3	37.5 (10.8, 74.9)	1.38 (0.27, 6.89)	0.69
Previously known diabetes mellitus
No	373	71.7 (66.0, 76.8)	147	28.3 (23.2, 34.0)	1
Yes	33	52.4 (36.7, 67.6)	30	47.6 (32.4, 63.3)	2.30 (1.22, 4.35)	0.01
Years of treatment for diabetes
≤5 years	20	60.6 (38.9, 78.8)		39.4 (21.2, 61.1)	1
6–10 years	7	41.2 (20.2, 65.9)	10	58.8 (34.1, 79.8)	2.19 (0.67, 7.18)	0.186
>10 years	6	46.1 (16.3, 79.1)	7	53.8 (20.9, 83.8)	1.79 (0.31, 10.12)	0.497
Previously known hypertension
No	338	73.8 (67.8, 79.0)	120	26.2 (21.0, 32.1)	1
Yes	68	54.4 (43.7, 64.7)	57	45.6 (35.3, 56.2)	2.36 (1.49, 3.73)	<0.001
Years of treatment for hypertension
≤5 years	30	55.6 (42.3, 68.0)	24	44.4 (31.9, 57.7)	1
6–10 years	16	55.2 (38.5, 70.8)	13	44.8 (29.2, 61.5)	1.01 (0.43, 2.35)	0.971
>10 years	7	41.2 (23.6, 61.3)	10	58.8 (38.7, 76.4)	1.78 (0.68, 4.63)	0.227
Prior heart disease
No	401	69.7 (63.7, 75.2)	174	30.3 (24.8, 36.3)	1
Yes	5	62.5 (28.0, 87.7)	3	37.5 (12.3, 72.0)	1.38 (0.31, 6.13)	0.667
Prior cerebrovascular accident
No	404	69.8 (63.8, 75.1)	175	30.2 (24.8, 36.2)	1
Yes	2	50 (19.6, 80.4)	2	50 (19.6, 80.4)	2.30 (0.55, 9.66)	0.249
Dyslipidemia
No	399	69.9 (63.7, 75.4)	172	30.1 (24.6, 36.3)	1
Yes	7	58.3 (35.9, 77.8)	5	41.7 (22.2, 64.1)	1.65 (0.61, 4.47)	0.316
History of snake bite
No	399	69.6 (63.7, 75.0)	174	30.4 (25.0, 36.3)	1
Yes	7	70 (38.4, 89.7)	3	30 (10.3, 61.6)	0.98 (0.26, 3.62)	0.979
History of Covid-19 infection
No	371	70.5 (64.3, 76.1)	155	29.5 (23.9, 35.7)	1
Yes	35	61.4 (43.6, 76.6)	22	38.6 (23.4, 56.4)	1.50 (0.70, 3.21)	0.289
History of smoking
Never	372	70.1 (63.8, 75.6)	159	29.9 (24.4, 36.1)	1
Past	20	66.7 (49.4, 80.4)	10	33.3 (19.6, 50.6)	1.16 (0.54, 2.49)	0.682
Present	14	63.6 (37.0, 83.9)	8	36.4 (16.1, 62.9)	1.33 (0.44, 3.97)	0.598
History of Smokeless tobacco use
No	385	70.8 (64.9, 76.1)	159	29.2 (23.9, 35.1)	1
Yes	21	53.8 (38.4, 68.6)	18	46.1 (31.4, 61.6)	2.07 (1.09, 3.94)	0.027
History of Alcohol consumption
No	367	70.8 (64.8, 76.3)	151	29.1 (23.7, 35.2)	1
Yes	39	60 (46.6, 72.1)	26	40 (27.9, 53.4)	1.62 (0.92, 2.84)	0.092
Type of diet
Vegetarian	95	74.2 (64.8, 81.8)	33	25.8 (18.2, 35.2)	1
Non-vegetarian	911	68.3 (61.6, 74.4)	144	31.6 (25.6, 38.4)	1.33 (0.81, 2.18)	0.253
Self-reported consumption of traditional medicines
No	405	69.8 (63.9, 75.1)	175	30.2 (24.9, 36.1)	1
Yes	1	33.3 (4.18, 85.2)	2	66.7 (14.8, 95.8)	4.6 (0.41, 51.54)	0.21
Self-reported consumption of over-the-counter analgesics
No	394	70.0 (63.9, 75.4)	169	30.0 (24.6, 36.1)	1
Yes	12	60 (36.6, 79.6)	8	40 (20.4, 63.4)	1.55 (0.58, 4.16)	0.377
BMI (Asian classification)
Underweight	44	68.7 (56.5, 78.9)	20	31.2 (21.1, 43.5)	0.87 (0.49, 1.52)	0.63
Normal weight	213	65.7 (58.5, 72.3)	111	34.3 (27.7, 41.5)	1
Over weight	117	77.0 (67.5, 84.3)	35	23.0 (15.7, 32.5)	0.57 (0.35, 0.91)	0.02
Obese	29	74.4 (58.6, 85.6)	10	25.6 (14.4, 41.4)	0.66 (0.31, 1.39)	0.277
Waist circumference
Normal	120	68.6 (59.9, 76.1)	52	31.4 (23.8, 40.1)	1
High	144	69.8 (60.4, 77.7)	66	30.2 (22.3, 39.6)	1.05 (0.65, 1.71)	0.818
Hemoglobin
>11 g/dL	292	75.3 (69.1, 80.5)	96	24.7 (19.4, 30.9)	1
<11 g/dL	114	58.8 (50.0, 67.0)	80	41.2 (33.0, 50.0)	2.13 (1.43, 3.18)	<0.001

Statistical test used: Logistic regression with significance at the level of 0.05.

The number of subjects with transient AKI, CKD and CKDu across the five agroclimatic zones is given in Supplement Table S4. The percentage of subjects with transient AKI has no uniform correlation with either CKD or CKDu across different agroclimatic zones. Further, the lower percentage of transient AKI compared to either CKD or CKDu in the north-east zone is unexplained.

UTCI and zonal difference

The modelled UTCI for the three months- July, August and December in 2023–across all zones in Fig. 6a showed UTCI >38 °C, an indicator of very strong heat stress¹⁷ was widely present in the north-east and south zones compared with the north-west zone during August, considered to be second summer in Tamil Nadu. The respective zonal prevalence of CKD were 7.68%, 6.4% and 2.16% and were statistically significant (p < 0.001). December was the coolest month of the calendar year, which is reflected in the same Fig. 6a. Further, the number of days with UTCI>38 °C, for more than 4 h in a single day in August was higher than in December and the lower in north-west zone compared with the north-east and south zones (Fig. 6b).

Fig. 6.

(a) UTCI daytime average by month across Tamil Nadu for July, August and December 2023. Grid spacing Grid spacing: 0.25° × 0.25°. Range (Max, min): (38.3, 21.8). (b) Days of the month with UTCI >38 °C for 4 h or more for July, August and December 2023. Grid spacing: 0.25° × 0.25°. Range (Max, min): (31, 0).

Diet and CKD

The overall percentage of participants on a vegetarian diet was 15.5% and the majority (67.2%) had animal protein intake on fewer days than two days per week. There was no significant association between diet and CKD (Table 2). However, both univariable and multivariable analyses showed a significant association between a vegetarian diet and CKDu (Table S2 and Fig. 5).

Discussion

The overall prevalence of CKD and CKDu among agricultural labourers was 5.31% (95% CI 4.58–6.13) and 2.66% (95% CI 2.14–3.26), respectively. The prevalence was significantly higher in the north-east region and southern region compared to the north-West region. Age, male sex, lack of formal education, diabetes, hypertension, history of high cholesterol, anaemia, exposure to smokeless tobacco and long hours of outdoor work were associated with CKD. Older age, male sex, lack of education, anaemia and long hours of outdoor work were associated with CKDu. The phenomenon of transient AKI was observed in 17.4%, and on comparing them with the CKD group, those with diabetes, hypertension, exposure to smokeless tobacco, anaemia and higher mean age and male sex a high likelihood of progression to CKD.

Prevalence of CKD

Although there have been many studies that have highlighted the presence of CKDu clusters in different parts of India as well as other parts of the world,^10,19,20 this is the first population-based study done across the entire state of Tamil Nadu, with robust methodology. Strategies to avoid bias included the use of stratified multi-stage cluster sampling and eGFR estimation performed twice, with a 3-month interval, using IDMS standards for creatinine measurement. CKDu was also defined as per the Sri Lankan Society of Nephrology guidelines. These reasons may account for the lower prevalence of CKD in our study compared to previous reports. A similar study done in Uddanam¹¹ found CKD in 10.24% of the participants, but the study was restricted to one geographical region that is rich in coconut and cashew plantations. Even in our study, the highest prevalence of CKD (7.7%) was in the north-east zone, where the cashew plantations are significant. The prevalence of CKD in our population is lower than the Mesoamerican and Indonesian hotspots.²¹

Education

Absence of formal education was strongly associated with CKD. The prevalence of CKD among graduates, as compared with individuals with no formal education, was 2.5% and 7.9% respectively. Education is an important surrogate of socioeconomic status, and often correlates with occupational heat stress exposure, limited access to health care, and unhealthy lifestyle and behaviour.²² In our study, 32.7% of the participants never went to school and 26.6% only had primary school education. Park and colleagues,²³ in their Mendelian randomisation study from the UK Biobank records, showed that individuals with fewer years of formal education (<16 years of education) had a higher prevalence of hypertension, diabetes, cerebrovascular diseases and CKD stages 3–5, than those with more years of education. The data from the PREVEND study²⁴ showed a similar association between low education and the prevalence of CKD and accelerated decline in eGFR. This is in contrast to an Indian study where diabetes and hypertension were more common among the educated urban population.²⁵

Age and comorbidities

Advancing age is a non-modifiable risk factor for both CKD and CKDu, with increasing prevalence after 53.5 years in our study, a finding that was also reflected in the SEEK-INDIA cohort²⁰ and the CARRS surveillance study.²⁶ Age had significant association even in Uddanam nephropathy,¹¹ it might reflect long years of exposure to both traditional and non-traditional risk factors like heat stress and agrochemical or environmental toxins. In contrast, CKD in the Mesoamerican region affected younger people.²⁷ The traditional risk factors, diabetes (aOR 2.42; 95% CI 1.56–3.75) and hypertension (aOR 3.17; 95% CI 2.18–4.60), contributed to nearly 50% of the burden of CKD even in agricultural workers, with the duration of diabetes and hypertension increasing the risk of CKD. The prevalence of diabetes and hypertension among the study population was 13.4% and 30.6%, quite similar to the National Family Health Survey in India conducted between 2015 and 2021.²⁸ Among those detected, two-thirds of patients with hypertension and 40% of those with diabetes, were diagnosed only during the study.

Smokeless tobacco

Smokeless tobacco (tobacco chewing) is a common high-risk behaviour among rural populations. According to the Global Adult Tobacco Survey-2 and the LASI study by Bharati and colleagues,²⁹ at a population level, 20.4% and 21.4% of adults respectively, were current users of smokeless tobacco. Individuals employed in agriculture were more likely to be users of smokeless tobacco. In the study by Sarker and colleagues³⁰ in Bangladesh, exposure to smokeless tobacco increased the risk of hypertension and CKD. In our study, the behaviour was found in 4.9% of participants with an OR of 2.36 and an aOR of 1.65. Compared to smoking, tobacco chewing has independent association with CKD in our study. Though there was no significant difference in the serum cotinine levels between smokers and tobacco chewers,³¹ the exposure to nicotine may be continuous in chewers compared to smokers. Nicotine in the tobacco causes glomerular hyperfiltration and vascular endothelial dysfunction leading to glomerulosclerosis and accelerated decline of GFR. Furthermore, smoking leads to continuous cadmium exposure, which is a tubulotoxin that leads to chronic irreversible damage to the tubulointerstitial compartment.³²

Heat stress and UTCI

Agriculture is one of the most common occupations associated with work exposure to heat stress. In our study, the median weekly hours of outdoor work was 35 (IQR 24–48). There was a significant difference in outdoor work hours between those with eGFR >60 mL/min per 1.73 m² (33.42 h) and those with CKD (36.18 h), with the difference increasing in participants diagnosed as CKDu (38.07 h). A Taiwanese nationwide population study among agricultural workers found that increased outdoor heat exposure was associated with higher risk of CKD by 2.3% per °C increase in ambient temperature.⁴ In the 7-year follow-up of young adults in Mesoamerican region, both outdoor work hours and exposure to a hot environment has significant association with loss of kidney function.³³ Although wet-bulb globe temperature (WBGT) was not measured in our study, we proceeded with UTCI modelling, which has a good correlation with WBGT³⁴ and showed very strong heat stress (>38 °C) in August in the north-east and southern zone and lesser heat stress in the north-west zone, and this was reflected in the prevalence of CKD and CKDu, which was significantly higher in the north-east zone. The advantage of the UTCI model is that it can be applied in a wider geographical region simultaneously, and data can be modelled for the entire calendar year and cumulative exposure to heat stress can be calculated. In our study, outdoor work of >24.5 h per week was found to be associated with CKD risk (aOR 1.78 [95% CI 1.17–2.72]).

Subclinical acute kidney injury

A phenomenon of subclinical AKI with eGFR <60 mL/min per 1.73 m² during Phase 1 was found in 406 participants which subsequently normalised in Phase 2. A similar phenomenon of acute Mesoamerican nephropathy in 247 Nicaraguan sugarcane workers was reported by Fischer RSB and colleagues²⁷ in which subjects were symptomatic with fever, body pain, nausea and vomiting and presented with AKI in a work place hospital, on subsequent follow-up 8.5% progressed to CKD. In our study none was significantly symptomatic, and compared with those with normal renal function, participants with reversible subclinical AKI were older, had poorer education, and were hypertensive. Probably, these risk factors play a role in the impairment of renal autoregulation and dysfunctional heat dissipation due to reduced skin blood flow and sluggish sweat glands³⁵ and those on antihypertensive drugs like angiotensin-converting enzyme inhibitors and beta-blockers have reduced heat tolerance,³⁶ which might lead to repeated subclinical AKI and progression to CKD as a result of continuous exposure to heat stress. Further, number of days with UTCI >38 °C for more than 4 h per day was higher in August compared to December. Our Phase 1 sampling was done in August and Phase 2 in December, the coolest month, which could be a reason for subclinical AKI. A study among California agricultural workers showed an elevation of core body temperature by > 1 °C in 36% of individuals studied, and AKI during the cross-shift in 14.9% of workers and was proportional to the workload.³⁷ Hsu RK and colleagues³⁸ documented clear evidence for transition of AKI to CKD and studies show even subclinical AKI had two–fold increased risk for future CKD.³⁹ Kupferman and colleagues⁴⁰ reported a similar phenomenon in sugarcane workers, where one-third of those workers with cross-shift AKI had sustained decline in GFR at one year of follow-up. A study among Nicaraguan sugarcane mill workers also showed that those with AKI had incomplete recovery in 50% of cases, with subsequent heat stress leading to repetitive AKI and progression to CKD over four years of follow-up.⁴¹ Water, rest and shade (WRS) intervention by Wegman and colleagues⁴² in cane-cutting workers who were provided 3 L of water, 10–15 min of rest every 1–1.5 h of work had less cross-harvest eGFR changes compared with the control group. NIOSH guidelines recommend 8 oz of water at 10–15 °C for every 20 min of work and both NIOSH and the European Agency for Safety and Health at Work have not given guidelines on permitted work hours in hot environments.⁴³ Future studies targeting workplace intervention strategies to mitigate the risk of AKI are needed.⁴⁴

Nutrition

Anaemia was found to be associated with CKD (aOR 2.89 [95% CI 1.99–4.18]), CKDu (aOR 2.28 [95% CI 1.32–3.93]) and progression of subclinical AKI to CKD (aOR 2.13 [95% CI 1.43–3.18]). Anaemia is an established risk factor in the progression of CKD, which causes hypoxia of renal tubular cells, contributing to chronic tubulointerstitial damage and oxidative stress.⁴⁵ Although association between vegetarian diet and CKDu (Table S2) as well as on comparing those with normal renal function with those who had transient AKI (Table S3) were statistically significant both in the univariable and multivariable analysis, in stratified analyses, dietary patterns (vegetarian, occasional non-vegetarian ≤2 days, or fully non-vegetarian >3 days) did not show statistically significant associations with CKDu prevalence within strata. Thus, diet alone may not be the primary determinant of CKDu risk.

Only 1.64% of the participants had symptoms attributable to volume overload in CKD; this suggests the need for proactive screening of the high-risk group. Unusually increased nocturnal frequency of micturition (more than two times a night) was present in 12.93% of participants, which points to the tubulointerstitial involvement in CKDu, with defective concentrating ability. However, the prevalence of nocturia in the CKDu and CKD was 14.6%, and among those with no kidney disease, 12.7%, with no significant difference. The symptom of nocturia and testing of early morning urine osmolality as an early marker of CKDu need further exploration in future studies.

Although eight studies have been done in India on the prevalence of CKD, all these studies are limited by their cross-sectional study design with either random or non-random sampling of participants. Serum creatinine was not re-measured after a 3-month interval to establish the diagnosis of CKD. Most of the studies were done in urban populations, with a heterogeneity in eGFR calculations because of varied estimating equations, and the use of non-standardised measurements of serum creatinine.¹⁹ This is the first study from India on the prevalence of CKD exclusively among agricultural workers across the entire state using stratified multi-stage cluster random sampling. There are some limitations to our study. There was no follow-up of participants with subclinical transient AKI or CKD. Renal ultrasound was not done in all participants diagnosed with CKD. Renal biopsies were not pursued in patients with normal-sized kidneys, and urine biomarkers and albumin-creatinine ratio were not analysed.⁹ Finally, since clinical history was collected through a questionnaire, there might have been some recall bias. This study was not explicitly designed to measure environmental exposures such as heat stress through WBGT or agrochemical use, which are key factors hypothesised to contribute to CKDu. Their exclusion limits the ability to establish causal pathways. Future studies should consider integrating such data and also need comparison of WBGT with UTCI in India to simplify the measurement of heat stress in field studies and to enhance the robustness of exposure–outcome relationships.

Conclusion

The prevalence of CKD in our study is lower than the global and Indian hotspots, with nearly half confined to the phenotype of CKDu. Low levels of education and absence of significant symptoms may delay self-referral and early diagnosis. Hence, health policies aimed at yearly measurement of serum creatinine and urine albumin-creatinine ratio among agricultural workers to identify hotspots of CKDu are needed. The phenomenon of reversible AKI needs further exploration in longitudinal studies across seasons for further evidence. Apart from WRS intervention in the workplace, interventional studies targeting daily time of exposure to hot environments based on UTCI are needed to guide work-hour regulation. Government policy measures that expand access to education and implementation of tobacco cessation programmes might help to mitigate the burden of CKD among agricultural workers.

Contributors

Conceptualisation–NG, SS, RS, TD, AS, STS.

Data curation–SS, NN, RS, SAR, JS, TML.

Formal analysis–SS, RS, TD, TML, JS, SAR, AIN.

Funding acquisition–SS, NG, AS, SR, STS, SJ.

Investigation–NG, SS, RS, TD, AS, NN, UM, SR.

Methodology–NG, SS, RS, TD, AS, SJ, MS, VR, RKM, AIN.

Project administration–SS, AS, STS, UM, MS, SR.

Resources–SS, AS, STS, SR.

Software–SS, SAR, JS.

Supervision–NG, SS, RS, TD, TML, VA.

Validation–NG, SS, RS, TD, AS, SJ, STS.

Writing–original draft–NG, SS, RS, TD, TML, VA.

Writing–review and editing–NG, SS, RS, TD, AS, SJ, VA, MS, VR, RKM, AIN.

Authors who accessed and verified the data—SS, RS, TML.

Authors responsible for the decision to submit the manuscript—SS, RS, NG.

Data sharing statement

The data that support the findings of this study are available from the corresponding author, NG, upon reasonable request, and can be shared with the permission of the funding agency.

Editor note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of interests

None.