Introduction
Over the past 3 decades, there has been an increasing recognition of a distinct subtype of chronic kidney disease (CKD) in specific rural communities, called CKD of uncertain etiology (CKDu). Regions affected by CKDu are characterized by high prevalence of advanced kidney disease or death from kidney disease.
Ongoing studies have focused on these regions and attempted to identify person-level risk factors for the development of kidney disease. However, these studies are limited by several factors. Nephrologists struggle to identify early kidney disease because of the marked variability of estimated glomerular filtration rate (eGFR) at levels > 60 ml/min per 1.73 m2, the internationally accepted threshold to identify “CKD”; and lack of more sensitive biomarkers. Second, despite the high relative incidence of meeting a threshold of eGFR < 60 ml/min per 1.73 m2, the absolute numbers of incident persons in prospective studies remains small.1
We tested the hypothesis that a CKDu hotspot would have a community-wide difference in eGFR distribution relative to the eGFR distribution in nonendemic communities by using data from 2 population-based studies in India.2,3
Results
A total of 1654 BLOOM participants and 10,362 UDAY participants had available data on serum creatinine (Supplementary Table S1 and Supplementary Figures S1 and S2). BLOOM participants were younger than UDAY participants (mean age: 29.1 [7.5] years vs. 47.6 (12.4) years in UDAY) (Table 1). There were proportionally more men in BLOOM (50.5%) than UDAY (45.3%). Diabetes prevalence was similar between the rural communities in UDAY (6.1% in North Rural and 5.7% in South Rural) and BLOOM (5.6%), but prevalence was higher in urban UDAY North and South (13.2% and 24.6%, respectively).
Table 1.
Baseline characteristics of participants in the UDAY and BLOOM studies
| Characteristics |
UDAY South |
UDAY North |
UDAY South |
UDAY North |
BLOOM South |
|---|---|---|---|---|---|
| Site | Urban, n (%) | Urban, n (%) | Rural, n (%) | Rural, n (%) | Rural, n (%) |
| Age categories | n = 2535 | n = 2672 | n = 2726 | n = 2429 | n = 1654 |
| 18–29 yrs | - | - | - | - | 1037 (62.7) |
| 30–44 yrs | 1239 (48.9) | 1150 (43.0) | 1210 (44.4) | 1033 (42.5) | 558 (33.7) |
| ≥ 45 yrs | 1296 (51.1) | 1522 (57.0) | 1516 (55.6) | 1396 (57.5) | 59 (3.6) |
| Missing | 0 | 0 | 0 | 0 | 0 |
| Mean age (SD), yrs | 46.8 (12.7) | 48.0 (12.0) | 46.9 (11.7) | 48.9 (13.2) | 29.1 (7.5) |
| Sex | |||||
| Male | 1174 (46.3) | 1273 (47.6) | 1252 (45.9) | 999 (41.1) | 836 (50.5) |
| Female | 1361 (53.7) | 1399 (52.4) | 1474 (54.1) | 1430 (58.9) | 818 (49.5) |
| Missing | 0 | 0 | 0 | 0 | 0 (0) |
| Education | |||||
| No formal education | 402 (15.9) | 540 (20.2) | 1674 (61.4) | 753 (31.0) | 221 (13.4) |
| Primary school | 782 (30.8) | 787 (29.5) | 800 (29.3) | 883 (30.9) | 123 (7.4) |
| Secondary school | 533 (21.0) | 732 (27.4) | 202 (7.4) | 658 (27.1) | 1013 (61.2) |
| Graduate school | 817 (32.2) | 613 (22.9) | 50 (1.8) | 135 (5.6) | 288 (17.4) |
| Missing | 1 (0.04) | 0 | 0 | 0 | 9 (0.5) |
| Occupation | |||||
| Agriculture | -- | -- | -- | -- | 905 (54.7) |
| Professional | 333 (13.1) | 388 (14.6) | 47 (1.8) | 113 (4.7) | -- |
| Skilled | 476 (18.8) | 440 (16.5) | 775 (28.4) | 383 (15.8) | 132 (8.0) |
| Semiskilled | 256 (10.1) | 238 (8.9) | 356 (13.1) | 219 (9.0) | -- |
| Unskilled work | 204 (8.0) | 105 (3.9) | 687 (25.2) | 114 (4.7) | 181 (10.9) |
| Unemployeda | 1266 (49.9) | 1501 (56.2) | 861 (31.6) | 1600 (65.9) | 430 (26.0) |
| Missing | 0 | 0 | 0 | 0 | 6 (0.4) |
| Diabetes | 624 (24.6) | 352 (13.2) | 156 (5.7) | 148 (6.1) | 93 (5.6) |
| Missing | 7 (0.3) | 0 | 3 (0.1) | 2 (0.1) | 0 |
| BP | |||||
| Systolic BP, mm Hg | 125.9 ± 21.1 | 127.8 ± 19.6 | 118.8 ± 20.7 | 124.0 ± 18.9 | 117.3 ± 13.1 |
| Diastolic BP, mm Hg | 75.7 ± 12.2 | 78.9 ± 11.1 | 70.5 ± 12.0 | 76.2 ± 11.2 | 72.9 ± 9.7 |
| Missing (%) | 5 (0.2) | 15 (0.6) | 14 (0.5) | 30 (1.2) | 263 (15.9) |
| Mean BMI (SD) | 25.8 (5.1) | 25.7 (5.7) | 21.5 (4.1) | 23.5 (5.7) | 23.2 (4.0) |
| Mean urine PCR (SD) | -- | -- | -- | -- | 0.11 (0.09) |
| Urine PCR categories | |||||
| 0–150 | -- | -- | -- | -- | 1280 (77.4) |
| 150–300 | -- | -- | -- | -- | 307 (18.9) |
| ≥ 300 | -- | -- | -- | -- | 63 (3.8) |
| Missing | -- | -- | -- | -- | 4 (0.2) |
| Mean UACR (SD) | 25.9 (184.0) | 23.2 (185.1) | 19.6 (163.4) | 13.7 (102.7) | - |
| UACR categories | |||||
| ≤ 30 | 2173 (85.7) | 2441 (91.4) | 2355 (86.4) | 2272 (93.5) | - |
| 30–300 | 193 (7.6) | 185 (6.9) | 137 (5.0) | 120 (4.9) | - |
| ≥ 300 | 32 (1.3) | 28 (1.0) | 21 (0.8) | 14 (0.6) | - |
| Missing | 137 (5.4) | 18 (0.7) | 213 (7.8) | 23 (0.9) | - |
BMI, body mass index; BP, blood pressure; PCR, protein-to-creatinine ratio; UACR, urine, albumin-to-creatinine ratio.
Number and proportion of unemployed who are female per site: 1026 (81.0) in UDAY South Urban, 1209 (80.5) in UDAY North Urban, 705 (81.9) in UDAY South Rural, 1209 (75.6) in UDAY North Rural, and 422 (98.1) in BLOOM.
Overall, BLOOM participants had lower median eGFR than UDAY participants across all sites. Median eGFR for BLOOM was 92.2 (25th, 75th percentile: 78.4, 105.00 ml/min per 1.73 m2 versus 104.5 (94.1, 113.6) for UDAY North Rural (Supplementary Table S2). Population eGFR distribution for BLOOM was systematically left-shifted compared with all UDAY sites for both age and sex strata (Figure 1).
Figure 1.
Age-stratified eGFR distribution by study sites. (a) Kernel density plot of eGFR distribution among all participants by each site in UDAY and BLOOM stratified by age. Relevant data are available in Supplementary Tables S2-S4. (b) Kernel density plot of eGFR distribution among men at each site stratified by age. Among the 30 to 44 years cohort, the median eGFR for BLOOM is lower than all UDAY sites. The median eGFR is 89.9 (25th, 75th percentile: 75.5, 101.2) ml/min per 1.73 m2 in BLOOM versus 112.7 (103.8, 118.0) ml/min per 1.73 m2 in UDAY North Rural (P < 0.001 for difference in distribution). Among the 45 years or older cohort, the median eGFR for BLOOM is 73.6 (58.7, 89.6) ml/min per 1.73 m2 in comparison with a median eGFR of 99.1 (86.1, 105.8) in UDAY North rural (P < 0.001). (c) Kernel density plot of eGFR distribution among women at each site stratified by age. Among the 30 to 44 years cohort, the median eGFR for BLOOM is lower than for other sites, with a median eGFR of 91.0 (25th, 75th percentile: 77.8, 102.1) in BLOOM versus 114.5 (106.4, 119.3) in UDAY North Rural (P < 0.001 for difference in distribution). Among the 45 years or older group, the sample size was limited for BLOOM (n = 7) but the median eGFR among BLOOM participants was lower than for UDAY North Rural participants. (d) Kernel density plot of eGFR distribution among all adults without CKD (defined as eGFR ≥ 60 ml/min per 1.73 m2) by each site in UDAY and BLOOM. Among the 30 to 44 years cohort, the median eGFR was lower for BLOOM than all other sites (P < 0.001 for differences in distribution). For example, whereas the median eGFR in BLOOM was 90.8 (25th, 75th percentile: 80.3, 101.7), it was 114.1 (105.7, 111.4) for UDAY North (P < 0.001 for difference in distribution). Among adults aged 45 years and older, median eGFR was not significantly different between BLOOM and other sites. CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate.
When stratified by sex, there was a 23 ml/min per 1.73 m2 difference between median eGFRs among men aged 30 to 44 years in BLOOM versus UDAY North Rural (Figure 1b; Supplementary Table S3). A similar (23 ml/min per 1.73 m2) difference in median eGFR was evident among women aged 30 to 44 years (Figure 1c; Supplementary Table S4).
Among participants without CKD (i.e., eGFR ≥ 60 ml/min per 1.73 m2), median eGFR for BLOOM was lower than all UDAY sites (Figure 1d). Similar patterns held for participants with and without diabetes (Supplementary Figure S3A and B). The proportion of patients with CKD (eGFR < 60 ml/min per 1.73 m2) was higher in BLOOM (6.6%) than in UDAY North Rural (2.0%) (Supplementary Table S5). Across all sites, older age, male sex, and higher body mass index were associated with lower eGFR (Supplementary Table S7).
Discussion
We find that participants of a community-based study residing in agricultural households in Andhra Pradesh, South India had systematically lower kidney function, as measured by calibrated creatinine-based eGFR, than participants of a second community-based study living in a rural area in North India and urban areas in North and South India. Unlike previous studies describing CKDu hotspots as primarily affecting men, we find a systematically left-shifted distribution of eGFR present among men and women, who belong to primarily agricultural communities in BLOOM. This large magnitude, community-specific difference could exist for several reasons, including pervasive nephrotoxic exposure, lower nephron endowment at birth, or remote exposure leading to kidney injury in childhood.1,4,5 Furthermore, communities with lower baseline eGFR are more vulnerable to subsequent kidney injury.
Our findings align with studies from pediatric populations in India and from Central America. Among 25,000 children across India, children living in rural regions consistently had lower eGFR than children from urban regions, with the highest prevalence of CKD (29.6%) in Andhra Pradesh.5 In our analysis, the median eGFR was lower among those aged 18 to 29 years in our CKDu-endemic (BLOOM) region, compared with older persons living in nonendemic areas, indicating lower baseline kidney function at a young age in a CKDu-endemic community.
Several reasons exist to focus on the distribution of eGFR rather than on a single threshold (i.e., eGFR < 60 ml/min per 1.73 m2). First, eGFR derives from serum creatinine-based formula, subject to variation due to differences in diet and body composition. Adding cystatin C to creatinine has improved correlation with measured GFR but remains subject to variations due to inflammation and diabetes.6,7 Second, disease presentation and progression may require multiple hits, with a predisposing factor amplified by autoimmune disease or viral infections, as has been seen for example with APOL1 nephropathy.8 Comparing kidney function distribution within populations with shared lifestyle and dietary habits therefore enables the detection of an “at risk community” and can signal the presence of a risk factor in subgroups less likely to manifest disease.
Correspondingly, although studies have described male predominance for the development of end-stage kidney disease in CKDu-endemic areas, we found a similarly left-shifted kidney function distribution among both women and men relative to North rural populations in India. Both sexes living in at-risk settings may share exposures to environmental toxins such as pesticides or heavy metals. However, women may be exposed indirectly (e.g., “a lower dose”) or may lack a “risk amplifier” (e.g., heat stress). In BLOOM, 52.1% of women were not working outside the home, suggesting that their route of exposure may be more indirect.
We also observed differences in the association of correlates of eGFR in our endemic and nonendemic regions, although this analysis is limited by sample size and differences in data collection (Supplementary Tables S6 and S7). Diabetes—defined in both cohorts using fasting glucose and medication use—was associated with higher eGFR in nonendemic UDAY sites, but with lower eGFR in BLOOM. This could reflect hyperfiltration, a phenomenon observed in many patients with type 2 diabetes.
The strengths of our study include the use of community-based random sampling, which reduces selection bias and use of an isotype dilution mass spectrometry-calibrated assay for serum creatinine-based eGFR which reduces measurement bias. Finally, by comparing eGFR distribution within Indian populations, we could focus on regional differences in kidney function. There are some limitations. Because the data are cross-sectional, we could not distinguish acute kidney injury from CKD. Enrollment criteria also differed between BLOOM and UDAY, with a higher proportion of younger participants and differences in ascertaining occupation in BLOOM. Given that UDAY and BLOOM were performed nearly a decade apart, there may be unmeasured confounders. UDAY excluded persons with renal failure. However, we would not expect this exclusion criteria alone to explain the observed shift in eGFR distribution because there was only 1 person with renal failure, defined as an eGFR < 15 ml/min per 1.73 m2, in BLOOM. Although previous data have shown that awareness and optimal treatment of CKD is rare in India, our study is nonetheless limited by lack of data on the use of renin-angiotensin-aldosterone axis inhibitors among participants with CKD.9
In summary, our community-level comparison of a CKDu-endemic and several nonendemic regions within India demonstrates a systematically lower eGFR in endemic areas. Using a strict threshold of eGFR < 60 ml/min per 1.73 m2 to define “significant” kidney disease may miss an already vulnerable population with an intrinsically lower eGFR. Community-level comparisons can facilitate earlier identification of disease and inform causal investigations. Comparisons of exposures and eGFR distribution between affected and unaffected communities could complement ongoing prospective efforts focused on persons within affected communities.1
Disclosure
All the authors declared no competing interests.
Acknowledgments
UDAY was supported by an unrestricted educational grant from the Eli Lilly and Company under the Lilly NCD Partnership Program. BLOOM is supported by the Medical Research Council/UK Research and Innovation (MR/T044527/1) and the Scottish Funding Council. PK had core funding from the UK Medical Research Council (MM-UU-00022/2) and the Scottish Government Chief Scientist Office (SPHSU17). NS received funding from the American Kidney Fund Clinical Scientist in Nephrology. SA, MR, and XY were supported by 5R01DK12713805. Funders were not involved with study design, collection, management, analysis, or writing this manuscript. We would like to thank the participants of both the UDAY and BLOOM studies. We would also like to acknowledge the contributions of Bharath Yandrapu, BLOOM Project Manager; Suresh Tulasi and Govinda Raju (BLOOM Field Supervisors); Ganga Rao, Lova Kumar, Kumari, Tirupathi, Ramesh, Sallamma, Basha, Raghavendra, Madhu, Rajesh, Ravi, Sreekala, and Kanakaraju (BLOOM Data Collectors); and Dr Rama (Qpath Laboratories).
Data Availability Statement
The data supporting this study comes from UDAY and BLOOM. Given participant privacy concerns, all requests for the data will be reviewed by the UDAY and BLOOM investigators and approval granted on a case-by-case basis.
Footnotes
Supplementary Methods.
Figure S1. Locations of BLOOM and UDAY studies.
Figure S2. Consort diagram for study analytical cohort (A) UDAY and (B) BLOOM.
Figure S3. (A and B) Age-stratified and condition-stratified eGFR distribution by study sites.
Table S1. Geographic differences between UDAY and BLOOM.
Table S2. Median eGFR in ml/min per 1.73m2 [25%, 75%] for all sites by age group.
Table S3. GFR distribution by age and location in men.
Table S4. GFR distribution by age and location in women.
Table S5. Proportion of participants with and without CKD (eGFR < 60 or ≥ 60).
Table S6. Multivariate logistic regression of correlates associated with CKD.
Table S7. Multivariate linear regression analyses of correlates associated with eGFR.
Supplementary Material
Supplementary Methods. Figure S1. Locations of BLOOM and UDAY studies. Figure S2. Consort diagram for study analytical cohort (A) UDAY and (B) BLOOM. Figure S3. (A and B) Age-stratified and condition-stratified eGFR distribution by study sites. Table S1. Geographic differences between UDAY and BLOOM. Table S2. Median eGFR in ml/min per 1.73m2 [25%, 75%] for all sites by age group. Table S3. GFR distribution by age and location in men. Table S4. GFR distribution by age and location in women. Table S5. Proportion of participants with and without CKD (eGFR < 60 or ≥ 60). Table S6. Multivariate logistic regression of correlates associated with CKD. Table S7. Multivariate linear regression analyses of correlates associated with eGFR.
References
- 1.Vlahos P., Schensul S.L., Anand S., et al. Water sources and kidney function: investigating chronic kidney disease of unknown etiology in a prospective study. npj Clean Water. 2021;4:50. doi: 10.1038/s41545-021-00141-2. [DOI] [Google Scholar]
- 2.Mohan S., Jarhyan P., Ghosh S., et al. UDAY: a comprehensive diabetes and hypertension prevention and management program in India. BMJ Open. 2018;8 doi: 10.1136/bmjopen-2017-015919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Jaacks L.M., Bliznashka L., Craig P., et al. Co-Benefits of Largescale Organic farming on huMan health (BLOOM): protocol for a cluster-randomised controlled evaluation of the Andhra Pradesh Community-managed Natural Farming programme in India. PLoS One. 2023;18 doi: 10.1371/journal.pone.0281677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hoy W.E., Hughson M.D., Singh G.R., Douglas-Denton R., Bertram J.F. Reduced nephron number and glomerulomegaly in Australian Aborigines: a group at high risk for renal disease and hypertension. Kidney Int. 2006;70:104–110. doi: 10.1038/sj.ki.5000397. [DOI] [PubMed] [Google Scholar]
- 5.Singh A., Verma M., Das M., Parija P.P., Nayak S., Jha V. Prevalence and associated factors of impaired kidney functions among children and adolescents in India: insights from the Comprehensive National Nutrition Survey (CNNS) 2016–18. BMC Pediatr. 2024;24:429. doi: 10.1186/s12887-024-04903-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Inker L.A., Eneanya N.D., Coresh J., et al. New creatinine- and cystatin C–based equations to estimate GFR without race. N Engl J Med. 2021;385:1737–1749. doi: 10.1056/NEJMoa2102953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chen D.C., Potok O.A., Rifkin D., Estrella M.M. Advantages, limitations, and clinical considerations in using cystatin C to estimate GFR. Kidney360. 2022;3:1807–1814. doi: 10.34067/KID.0003202022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Daneshpajouhnejad P., Kopp J.B., Winkler C.A., Rosenberg A.Z. The evolving story of apolipoprotein L1 nephropathy: the end of the beginning. Nat Rev Nephrol. 2022;18:307–320. doi: 10.1038/s41581-022-00538-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tonelli M., Tiv S., Anand S., et al. Diagnostic yield of population-based screening for chronic kidney disease in low-income, middle-income, and high-income countries. JAMA Netw Open. 2021;4 doi: 10.1001/jamanetworkopen.2021.27396. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Methods. Figure S1. Locations of BLOOM and UDAY studies. Figure S2. Consort diagram for study analytical cohort (A) UDAY and (B) BLOOM. Figure S3. (A and B) Age-stratified and condition-stratified eGFR distribution by study sites. Table S1. Geographic differences between UDAY and BLOOM. Table S2. Median eGFR in ml/min per 1.73m2 [25%, 75%] for all sites by age group. Table S3. GFR distribution by age and location in men. Table S4. GFR distribution by age and location in women. Table S5. Proportion of participants with and without CKD (eGFR < 60 or ≥ 60). Table S6. Multivariate logistic regression of correlates associated with CKD. Table S7. Multivariate linear regression analyses of correlates associated with eGFR.
Data Availability Statement
The data supporting this study comes from UDAY and BLOOM. Given participant privacy concerns, all requests for the data will be reviewed by the UDAY and BLOOM investigators and approval granted on a case-by-case basis.