Epidemiology of chronic kidney disease in young patients in developing countries

Abstract

BACKGROUND

Chronic kidney disease (CKD) is an important but under-recognized cause of morbidity in young people living in low- and middle-income countries (LMICs), where epidemiological data in this age group are limited and fragmented. LMICs, as defined by the World Bank, have a gross national income per capita between United States dollar 1136 and United States dollar 13845 (2024).

AIM

To summarize the prevalence of CKD among individuals aged ≤ 25 years in LMICs and to explore sources of between-study variability.

METHODS

We performed a systematic review and meta-analysis following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines. PubMed, EMBASE, Cochrane Library, and Google Scholar were searched for original studies reporting CKD prevalence in LMIC populations aged 0-25 years. Study-level proportions were pooled using a random intercept logistic regression model (generalized linear mixed model) with logit transformation. Heterogeneity was quantified using τ², I², and Q statistics. Prespecified subgroup analyses stratified studies by world region and study setting (community-based, hospital/clinic-based, CKD/end-stage renal disease registries, and specific disease cohorts). Random-effects meta-regression examined the contribution of region and setting to heterogeneity. Sensitivity analyses excluding registry and high-risk cohorts were undertaken to approximate prevalence in more general populations.

RESULTS

Nineteen studies from 17 countries (339940 participants; 96674 CKD cases) were included. The overall pooled prevalence was 11.7% (95% confidence interval: 5.5%-23.3%), with a prediction interval spanning 0.3%-84% and extreme heterogeneity (τ² = 3.41; I² = 99.9%). Subgroup analyses showed a clear gradient by setting, from lowest prevalence in community-based samples to highest in registries and disease-specific cohorts. Meta-regression indicated that setting explained 36.8% and region 28.2% of between-study variance, while both combined explained 61.4%. In sensitivity analyses restricted to more general populations, pooled prevalence was in the 4%-5% range (random-effects 95% confidence interval: 2%-9%), although heterogeneity remained high.

CONCLUSION

CKD in young people in LMICs is common and highly variable, largely reflecting differences in study setting and geography, and warrants targeted early detection and surveillance strategies.

Key Words: Kidney failure; Chronic/epidemiology; Developing countries; Adolescent; Child; Young adult

Core Tip: Chronic kidney disease (CKD) in 0-25-year-old patients living in low- and middle-income countries is substantially under-recognized and highly variable across settings. This systematic review and meta-analysis show that CKD prevalence in individuals ≤ 25 years can reach clinically important levels even in community-based samples, and that much of the between-study variability is explained by study setting and region. These findings highlight the urgent need for standardized screening strategies, harmonized epidemiological methods, and national CKD registries focused on young populations in resource-limited settings.

INTRODUCTION

Chronic kidney disease (CKD) is a progressive condition defined by a reduction in kidney function lasting at least three months, identified by persistently elevated serum creatinine, reduced estimated glomerular filtration rate (eGFR), and/or albuminuria[1-3]. eGFR is a lab-calculated measure of kidney function, derived from factors like age, sex, ethnicity, and blood creatinine levels. A normal eGFR is typically above 90 mL/minute/1.73 m², reflecting healthy kidneys. In many developing countries, early-stage CKD remains undetected because of limited awareness, socioeconomic barriers, and restricted access to diagnostic testing[2]. CKD multiplies cardiovascular risk by 2-fold to 4-fold and imposes a substantial financial burden, particularly in low- and middle-income countries (LMICs) where health systems are already constrained[3-5]. Disease onset and progression are further driven by nephrotoxic medications, environmental pollutants, and structural inequities in healthcare access[6,7]. Addressing this growing burden requires international collaboration, regular screening, public education, and robust epidemiological research, particularly in underserved regions such as South Asia[7-11]. Early detection in younger populations in LMICs is especially important, as it offers a critical opportunity to reduce long-term morbidity and premature mortality[12].

Pediatric CKD differs substantially from adult CKD in both aetiology and management. In children, CKD is most commonly due to congenital anomalies of the kidney and urinary tract, progression from acute kidney injury, schistosomiasis, and other structural or infectious causes, often requiring close collaboration between nephrologists and urologists[13]. Hereditary factors account for up to 20% of congenital anomalies of the kidney and urinary tract, with single-gene mutations explaining around 10% of sporadic cases and copy-number variants a further 15%, while approximately 75% of cases still lack an identifiable genetic cause. Epigenetic mechanisms, including histone modifications and DNA methylation pathways, influence nephron development in response to fetal energy supply, and apolipoprotein L1 risk variants in some populations of African ancestry increase susceptibility to kidney injury and accelerate CKD progression[14].

These biological vulnerabilities intersect with infectious and non-communicable comorbidities. Kidney disease is a major complication of human immunodeficiency virus (HIV) infection; studies from sub-Saharan Africa report CKD prevalence in children living with HIV ranging from 0% to 31.6%[15]. In young patients, diabetes mellitus (type 1 and type 2), hypertension, and glomerulonephritis are dominant causes of CKD, whereas in young children, incomplete nephron development and congenital anomalies predominate, leading to growth impairment and neurocognitive consequences if not recognized and managed early[16]. Taken together, these factors suggest that CKD in young people in LMICs is likely both underdiagnosed and unevenly distributed across settings and regions, with available data fragmented, inconsistently defined, and rarely focused specifically on individuals aged ≤ 25 years. Against this background, the present systematic review and meta-analysis aimed to estimate the prevalence of CKD in young people living in LMICs, to describe how this prevalence varies across countries, world regions, and study settings, and to provide an updated synthesis that can inform future research, surveillance strategies, and policy initiatives targeting this vulnerable age group.

MATERIALS AND METHODS

This meta-analysis was conducted to estimate the pooled prevalence of CKD among young individuals aged 25 years or younger living in LMICs. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines[17]. and the protocol was prospectively registered in the International Prospective Register of Systematic Reviews (No. CRD420250650310, available from: https://www.crd.york.ac.uk/PROSPERO/view/CRD420250650310) to ensure methodological transparency and reproducibility. This age group was selected a priori as a clinically important yet under-represented population in LMICs, where constrained health systems and underreporting of chronic conditions may delay diagnosis and treatment.

A comprehensive literature search was undertaken in PubMed/MEDLINE, EMBASE, Cochrane Central, and Google Scholar to identify original studies reporting the prevalence of CKD among individuals aged 0-25 years in LMIC settings. English was used as the only language, which was a barrier to reaching articles written in other languages. Unpublished or grey literature was not handled, as it may affect the validity of data in the reports. The search strategy combined Medical Subject Headings (MeSH) and appropriate Boolean operators: [“Kidney Diseases/epidemiology” (Mesh)] OR “Chronic Kidney Disease” (Title/Abstract) OR “CKD” (Title/Abstract) OR “Renal Failure” (Title/Abstract) OR “Chronic Renal Disease” (Title/Abstract) OR “Kidney Failure” (Title/Abstract) AND [“Developing Countries” (Mesh) OR “Low- and Middle-Income Countries” (Title/Abstract) OR “LMICs” (Title/Abstract)] AND [“Pediatric” (Title/Abstract) OR “Children” (Title/Abstract) OR “Young Patients” (Title/Abstract)]. The literature search was carried out between April 2025 and May 2025, with no restrictions on publication date. Mendeley software was used for reference management, organization of citations, and support during data extraction. In addition, the reference lists of all included studies were manually screened to identify further eligible articles, in order to maximize the retrieval of relevant evidence from the published literature.

Inclusion criteria and exclusion criteria

Studies were eligible if they reported original, peer-reviewed data on CKD prevalence among individuals aged 0-25 years living in LMICs, as defined by the World Bank (low- and middle-income economies based ongross national income per capita (gross national income earned is the proportion of the total income earned either in the country or abroad by residents divided by the total population)[18]. The age group 0-25 years was chosen to capture the set of populations that represent a critical window for early-onset CKD, with long-term public health implications, as well as the aetiology, GFR trajectories, and long-term complication risk differ substantially from those of young patients. Eligible studies had to define CKD according to internationally accepted criteria (e.g., Kidney Disease: Improving Global Outcomes, Kidney Disease Outcomes Quality Initiative) and provide sufficient data to calculate prevalence. Studies were excluded if they: (1) Focused on people whose ages are < 25 years or not residing in LMICs; (2) Included only acute kidney injury or other non-CKD renal disorders; (3) Were systematic reviews, editorials, commentaries, or case series without prevalence data; or (4) Did not apply standard CKD definitions or lacked adequate methodological detail.

Study selection and data extraction

The study selection process followed PRISMA 2020 recommendations. After removal of duplicates, titles and abstracts were screened to exclude clearly irrelevant records. Full-text articles were then assessed against the predefined eligibility criteria, and only studies reporting original prevalence data for the target age group in LMIC settings were retained. Reference lists of included papers were manually searched to identify additional eligible studies. Data were extracted using a standardized form, including: (1) Study identifiers (first author, year, country); (2) Total sample size and number of CKD cases; (3) Age range and CKD definition; (4) Contextual indicators such as gross domestic product and healthcare access index, when available; and (5) Methods used for estimating kidney function. All entries were cross-checked for internal consistency, and CKD staging was aligned with Kidney Disease Outcomes Quality Initiative criteria where possible. Methodological quality was assessed using the Joanna Briggs Institute checklist for prevalence studies. The overall selection and extraction strategy was guided by a Population/Intervention/Comparison/Outcome framework (young individuals ≤ 25 years in LMICs with CKD as the condition of interest and prevalence as the primary outcome), ensuring consistent application of inclusion criteria and transparent reporting.

Statistical analysis

For each study, CKD prevalence was calculated as the proportion of affected participants in the total sample, with 95% confidence interval (CI) derived using the Clopper-Pearson exact method. Study-level prevalence was pooled using a random-intercept logistic regression model (generalized linear mixed model) with a logit transformation of proportions (PLOGIT), as implemented in the meta prop function of the meta package. Between-study heterogeneity was quantified using the maximum-likelihood estimator for τ², together with I², H, and Q statistics. Pooled logit estimates and their confidence limits were back-transformed to the proportion scale using the inverse logit function, and 95% prediction intervals were obtained from the fitted mixed-effects model.

To explore sources of heterogeneity, we conducted prespecified subgroup analyses by world region (Africa, Asia, Latin America, and other) and by study setting (community-based, hospital/clinic-based, CKD/end-stage renal disease (ESRD) registries, and specific disease cohorts). Mixed-effects meta-regression models on the logit-transformed prevalence were then fitted to examine the contribution of region and setting, separately and combined, to between-study variance, with pseudo-R² estimated from the relative reduction in τ². Sensitivity analyses repeated the main model after excluding CKD/ESRD registries, disease-specific cohorts, and extreme hospital-based samples, to approximate prevalence in more general young populations. All analyses were performed in R (version 4.4.2) using the meta and metafor packages.

RESULTS

Study selection

The study selection process, summarized in Figure 1, followed PRISMA 2020 guidelines and combined studies from a previous version of the review (n = 172) with newly identified records. Overall, 170 records were retrieved from electronic databases, 5 from trial registers, and 22 from grey literature sources such as websites. After removal of duplicate records (n = 2) and exclusion of clearly ineligible or irrelevant entries (n = 25), 175 records underwent title and abstract screening, of which 100 were excluded. Seventy-five full-text reports were sought for retrieval, but 22 could not be obtained. Of the 53 full texts assessed for eligibility, 18 were excluded based on predefined criteria (inappropriate population, setting, or outcome). Ten additional studies met the inclusion criteria and were added to those from the previous version, resulting in a total of 19 studies included in the final quantitative synthesis and meta-analysis. These 19 studies encompassed diverse populations across 17 LMICs and together included 339940 individuals aged 25 years or younger, of whom 96674 were diagnosed with CKD. Sample sizes varied markedly, ranging from 73 participants in the smallest cohort[19] to over 212000 individuals in the Nigerian study by Ulasi and Ijoma[20]. Most studies were conducted in Africa, Asia, and Latin America, reflecting the geographical focus of CKD research in developing settings. A detailed description of each study - including first author, year of publication, country, total sample size, and number of CKD cases (Table 1).

Figure 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 flow diagram of the systematic review and meta-analysis.

Table 1 Summary of included studies.

Ref.	Country	Setting and design	Patients	Intervention	Comparison	Outcome
Abraham et al[22], 2016	India	Prevalence study. South Asia, including India, Bangladesh, Pakistan, Nepal, Bhutan, and Sri Lanka and Afghanistan	132	Screening	N/A	112
Degenaar et al[23], 2003	South Africa	A cross-sectional study with participants recruited from Potchefstroom and its surroundings	956	Questionnaires were used based on demographics, anthropometric, and some other measurements, such as biomarkers, to stratify the population	N/A	Individuals with low eGFR, low uromodulin, and high CKD273 classifier levels exhibited unfavorable cardiovascular profiles; in adjusted analyses, eGFR was inversely associated with HDL-C and GGT, while the CKD273 classifier was positively associated with age, HDL-C, and GGT
Garcia-Garcia et al[9], 2020	Mexico	Cross-sectional prevalence comparison study of Poncitlan to Jalisco in Mexico	51789 in Jalisco and 427 in Poncitlan	Data were collected prospectively in Jalisco using standardized forms on participants aged 2 years or older, and fasting blood glucose levels were measured; blood specimens of participants were used to measure serum creatinine and glucose; urine specimens for dipstick urinalysis, adult hypertension, diabetes mellitus, BMI, percentiles of children calculated > 95th percentile or < 5th percentile classified as overweight or malnourished, respectively		50626; the prevalence of CKD and proteinuria in the studied group was similar to that of the overall Mexican population; in Poncitlan, young patients were more likely to be female but of comparable age compared to other areas; children in Poncitlan were younger but had a similar gender distribution; notably, both CKD and proteinuria rates were significantly higher among young patients from Poncitlan compared to other municipalities; proteinuria was also much more common among children in Poncitlan than among their peers from other areas
Hafez et al[10], 2019	Egypt	Assessed prevalence rate, etiology, and other risk factors for end-stage renal disease (ESRD) in Aswan governorate in upper Egypt, during the period from June 2017 to June 2018	1000	Patients with end-stage renal disease on maintenance hemodialysis; this cross-sectional study was carried out on 1000 patients with ESRD		1000, the leading causes of ESRD in Aswan Governorate are hypertensive renal disease, diabetic nephropathy, and obstructive uropathy; other causes include glomerulonephritis, chronic pyelonephritis, congenital conditions, analgesic nephropathy, preeclampsia, polycystic kidney disease, gouty nephropathy, and lupus nephritis, with some of the causes being unknown
Zekry et al[24], 2023	Egypt	This study included 1000 hemodialysis patients aged 18 to 60 from Fayoum, Egypt	1000	Participants were recruited from local hospitals, and data were gathered from medical records and, when needed, directly from patients or their relatives		1000; the main causes of ESRD were hypertension, kidney stones, unknown causes, and both hypertension and diabetes; significant gender differences were found in the frequency of original kidney diseases
Gautam et al[25], 2024	Nepal	A retrospective prevalence study was conducted from 2022 to 2023 in the pediatric department	1116	This study examines kidney disease patterns in children admitted to the hospital over seven years, focusing on treatments given and patient outcomes	N/A	63 (5.6%); most affected were school-age children, while newborns and infants had the fewest cases; urinary tract infection was the most common condition, followed by acute kidney injury; treatments included medications, dialysis, and surgery; most patients improved and were discharged to the Nephrology clinic; some left against medical advice, died, or were referred elsewhere
Hamilton et al[26], 2020	Malawi	The cross-sectional study from January-August 2018 collected bio samples and anthropometric data in two Malawian population settings	821	Study analyzed data from 821 healthy young patients in Malawi to assess kidney function and its risk factors; researchers collected biological and body measurements and used the CKD-EPI equation to estimate glomerular filtration rate (eGFR); regression models were used to identify factors associated with reduced eGFR	N/A	783
Ulasi and Ijoma[20], 2010	Nigeria	ESRD patients seen at the University of Nigeria Teaching Hospital, Enugu, South-East Nigeria	1538	ESRD patients seen were recruited; records from the accident and emergency department, medical-out-patients, wards, and dialysis unit were used	N/A	908; the outlook for CKD patients in Nigeria is poor due to limited access to renal-replacement therapy, mainly because of high costs; this highlights the importance of prevention to lessen CKD’s impact
Khan et al[27], 1996	Albania	Incidence and outcome study of patients on the need for renal services in Albania	84	A review of case notes for all patients in Tirana who had a serum creatinine concentration of 300 μmol/L or higher during 1992, with their outcomes tracked over two years	The serum creatinine concentration of 300 μmol/L or higher during 1992 was reviewed	65
Gheissari et al[28], 2012	Iran	This study retrospectively analyzed the medical records of children under 19 years old who were hospitalized for CKD at St Alzahra Hospital in Isfahan, Iran, from November 2001 to December 2011	268	The study found that most children with CKD presented with advanced disease, and the majority required dialysis; glomerular diseases were the main cause, and anemia was very common; the transplantation rate was low, and younger children faced a higher risk of death; the incidence and mortality rates of advanced CKD were significant in the population studied	N/A	144
Priyanka et al[19], 2024	India	Prospective cohort study of children aged 1-18 years with biopsy-proven primary FSGS followed from January 2010 to June 2023 in a tertiary-care center in India	73	Children aged 1-18 years with biopsy-proven primary FSGS were enrolled, and their clinical profile, histological characteristics, kidney outcomes, and predictors of adverse outcomes were determined	N/A	52
Halle et al[29], 2009	Cameroon	Medical files of patients received at the outpatient department of nephrology from January 2001 to December 2003 at the Yaounde General Hospital were reviewed	183	Data on etiologic and co-morbidity of CRF recorded	N/A	160
Orta-Sibu et al[30], 2002	Venezuela	A report epidemiological data on renal disorders in children in Venezuela from 14 centers from January 1998 to December 1998	3624.0	Data were collected by chart review in all patients; because some patients are seen first in a small center near their home and then sent to a pediatric nephrology center for further investigations, the name and date of birth were checked in order to avoid double registry	N/A	2031
Wong et al[31], 2015	Cambodia	A retrospective observational study with data collection and analysis of patient records in the Mercy Medical Center (MMC) in Cambodia from April to December 2012	915	Between April and December 2012, 1013 new patient records at Mercy Medical Center in Cambodia were reviewed; of these, 915 adult patients were analyzed; patients with a history of hypertension were excluded from blood pressure analysis, and those with diabetes, hypertension, chronic kidney disease, or renal symptoms were excluded from urinalysis analysis	N/A	820; out of 820 patients without a history of hypertension, 8.9% had abnormal blood pressure, with 7.3% meeting the criteria for high blood pressure and 1.6% having isolated systolic hypertension; additionally, 30.6% showed significant proteinuria or hematuria, and 39.0% had elevated urine white blood cell counts; overall, 53.9% of these patients had at least one urinary abnormality
Plata et al[32], 1998	Bolivia	Educational renal campaign conducted in three selected areas	14082	Fresh urine samples were tested with a dipstick and examined under a microscope; patients with abnormal results received additional tests to confirm diagnoses and were followed for three years	N/A	4261; among 820 patients without hypertension, 8.9% had abnormal blood pressure; over half showed at least one abnormal urinalysis result, with 7.3% having high blood pressure, 1.6% isolated systolic hypertension, 30.6% significant proteinuria or hematuria, and 39% elevated urine white blood cells
Al-Rasheed et al[33], 1996	Saudi Arabia; the classification of Saudi Arabia as “developing country” is debatable and may not fully match current WB categories	Between April 1982 and September 1994, renal biopsies were performed in children at King Khalid University Hospital, Riyadh, Saudi Arabia	167	The reports the distribution of kidney disease types: 23.3% had minimal change lesions, 24% had mesangial proliferative glomerulonephritis, and 24% had focal segmental glomerulosclerosis; there was a higher incidence of congenital nephrotic syndrome and Alport’s syndrome compared to Western countries, while IgA nephropathy was rare (3%), and type II membranoproliferative glomerulonephritis was absent	N/A	156
Lou-Meda[34], 2015	Guatemala	A recent analysis of the Guatemalan Registry of Pediatric CKD included patient records from May 2004 to April 2013	432	Those classified as having CKD stage 2 or greater based on the most recent available estimated glomerular filtration rate (eGFR)	N/A	193
Ndongo et al[35], 2024	Senegal	A cross-sectional study in three northern regions in Senegal with a two-stage cluster sampling	2441	Non-pregnant women, recent hospital patients, those with recent symptoms, and people on renal therapy; data was gathered at participants’ homes using a modified WHO questionnaire; blood samples were analyzed for creatinine, lipids, and sugar; the estimated glomerular filtration rate was calculated using the CKD-EPI 2021 formula; the sample size accounted for a 5% precision, 80% power, and a 10% attrition rate	N/A	High blood pressure was found in 52% and CKD in 17.8% of participants; three out of five people with HBP were previously undiagnosed; CKD was most common among those with known HBP (30.5%), less common in those with undiagnosed HBP (19.1%), and least common in people with normal blood pressure (10.9%); older age and female sex were linked to a higher risk of CKD
Zhang et al[36], 2008	China	Cross-sectional population-level study	13925	13925 representative sample among 124410000 people living in Beijing		1% (1430000) with CKD

N/A: Not applicable; eGFR: Estimated glomerular filtration rate; CKD: Chronic kidney disease; HDL-C: High density lipoprotein cholesterol; GGT: Gamma-glutamyltransferase; ESRD: End-stage kidney disease; CKD-EPT: Chronic kidney disease epidemiology collaboration; FSGS: Focal segmental glomerulosclerosis; CFR: Case fatality rate; MMC: Mercy medical center; WB: World bank; IgA: Immunoglobulin A; WHO: World Health Organization.

Using the random-effects generalized linear mixed model, the pooled prevalence of CKD among young individuals in developing countries was estimated at 11.55% (95%CI: 5.36%-23.16%), reflecting substantial variability and statistical uncertainty (Figure 2). Prevalence estimates ranged from 0.16% in a Bolivian community survey to 84.85% in a small high-risk Indian cohort, underscoring the influence of differences in study setting, population risk profile, diagnostic criteria, and healthcare access across countries (Table 2). The corresponding 95% prediction interval was very wide, indicating that prevalence in a new, similar study could plausibly fall anywhere within this range (Figure 3A). Between-study heterogeneity was extreme (τ² = 3.41; I² = 99.9%; Q = 20360.7; P < 0.001), consistent with the broad spread of individual study estimates (Figure 3B).

Figure 2 Forest plot of chronic kidney disease prevalence among young individuals (≤ 25 years) in low- and middle-income countries from all 19 included studies, showing study-specific proportions with 95% confidence intervals and the pooled common- and random-effects estimates from the generalized linear mixed model. CI: Confidence interval.

Figure 3 Funnel plot and generalized linear mixed model meta-analysis of chronic kidney disease. A: Funnel plot of logit-transformed chronic kidney disease (CKD) prevalence with pseudo 95% confidence limits. Funnel plot displaying the standard error (Y-axis) against the logit-transformed prevalence estimates from each study (X-axis) under the random-intercept logistic regression model. The vertical dotted line represents the pooled logit prevalence, and the dashed lines indicate pseudo 95% confidence limits. The observed asymmetry reflects the marked between-study heterogeneity and suggests potential small-study or setting-related effects rather than a simple pattern of publication bias; B: Influence diagnostics for the generalized linear mixed model meta-analysis of CKD prevalence in young people in low- and middle-income countries. Influence plot showing each included study (labelled by study ID) according to its squared Pearson residual (X-axis) and its influence on the pooled prevalence estimate (Y-axis), under the random-intercept logistic regression model. Studies in the upper-right quadrant (notably studies 14 and 19) have both large residuals and greater influence on the overall result, indicating potentially influential outliers that contribute disproportionately to the observed heterogeneity.

Table 2 Individual study prevalence of chronic kidney disease.

Ref.	Country	Total patients	Number of CKD cases	Prevalence (95%CI)
Degenaar et al[23], 2003	South Africa	956	240	0.2510 (0.2238-0.2798)
Garcia-Garcia et al[9], 2020	Mexico	52216	5484	0.1050 (0.1024-0.1077)
Hafez et al[10], 2019	Egypt	1,000	53	0.0530 (0.0399-0.0688)
Zekry et al[24], 2023	Egypt	1000	103	0.1030 (0.0849-0.1235)
Gautam et al[25], 2024	Nepal	116	6	0.0517 (0.0192-0.1092)
Hamilton et al[26], 2020	Malawi	821	36	0.0438 (0.0309-0.0602)
Ulasi and Ijoma[20], 2010	Nigeria	212673	79862	0.3755 (0.3735-0.3776)
Khan et al[27], 1996	Albania	84	8	0.0952 (0.0420-0.1791)
Gheissari et al[28], 2012	Iran	653	181	0.2772 (0.2432-0.3132)
Priyanka et al[19], 2024	India	73	21	0.2877 (0.1877-0.4055)
Halle et al[29], 2009	Cameroon	140	23	0.1643 (0.1071-0.2362)
Orta-Sibu et al[30], 2002	Venezuela	3624	59	0.0163 (0.0124-0.0210)
Wong et al[31], 2015	Cambodia	1013	7	0.0069 (0.0028- 0.0142)
Plata et al[32], 1998	Bolivia	14082	22	0.0016 (0.0010-0.0024)
Al-Rasheed et al[33], 1996	Saudi Arabia	167	12	0.0719 (0.0377-0.1222)
Lou-Meda[34], 2015	Guatemala	1545	432	0.2796 (0.2573-0.3027)
Ndongo et al[35], 2024	Senegal	2441	1704	0.6981 (0.6794-0.7162)
Zhang et al[36], 2008	China	8301	47204	0.18 (0.17-0.18)
Abraham et al[22], 2016	India	132	112	0.8485 (0.7757- 0.9049)

CKD: Chronic kidney disease; CI: Confidence interval.

Given this heterogeneity, we conducted a series of sensitivity analyses to approximate CKD prevalence in more general populations. First, we excluded seven studies that predominantly represented high-risk clinical populations (CKD registries, specific disease cohorts, or extreme hospital-based samples), leaving 12 studies with lower prevalence and 121467 participants. In this restricted set, the random-effects pooled prevalence fell to 4.4% (95%CI: 2.1%-9.3%), although heterogeneity remained very high (τ² = 1.91; I² = 99.5%; Q = 2,014.7; P < 0.001) (Figure 4A). A further restriction to 10 studies with a prevalence considered to be closest to community or mixed background samples yielded a pooled prevalence of 3.3% (95%CI: 1.5%-7.3%) in 74123 participants, again with substantial heterogeneity (τ² = 1.76; I² = 98.8%; Q = 732.5; P < 0.001) (Figure 4B). Together, these analyses suggest that prevalence is appreciably lower in more general populations than in high-risk clinical cohorts, but still highly variable across settings.

Figure 4 Sensitivity analysis forest plot. A: Sensitivity analysis forest plot of chronic kidney disease prevalence restricted to 12 studies with prevalence ≤ 0.20, excluding the highest-risk clinical and registry cohorts; study-specific estimates and pooled common- and random-effects prevalences are shown, with heterogeneity statistics reported below; B: Further sensitivity analysis forest plot of chronic kidney disease prevalence restricted to 10 studies with prevalence ≤ 0.16, approximating more general young populations; individual study proportions, pooled estimates and residual between-study heterogeneity from the generalized linear mixed model are displayed. CI: Confidence interval.

Subgroup analyses by world region, performed in the full dataset and in both restricted datasets after excluding studies with prevalence > 0.20 and prevalence > 0.16, consistently showed marked differences in pooled prevalence between regions (Figures 5 and 6). In the full 19-study dataset, random-effects estimates were highest in Africa (4 studies; pooled prevalence 35.8%, 95%CI: 18.4%-57.8%), followed by Asia (7 studies; 15.0%, 95%CI: 4.0%-42.5%), with lower pooled prevalence in Latin America (4 studies; 3.2%, 95%CI: 0.4%-20.4%) and in the group classified as other regions (4 studies; 6.7%, 95%CI: 4.6%-9.7%). The test for subgroup differences under the random-effects model was statistically significant (Q between = 18.2; df = 3; P = 0.0004), indicating that geography contributed meaningfully to variation in reported prevalence across studies.

Figure 5 Forest plot of chronic kidney disease (≤ 25 years). A: Forest plot of chronic kidney disease prevalence in young people (≤ 25 years) stratified by world region for all 19 included studies. Pooled proportions are shown for each region (Africa, Latin America, Other, and Asia) under common-effect and random-effects models, together with overall pooled estimates and heterogeneity and subgroup-difference statistics; B: Forest plot of chronic kidney disease prevalence in young people stratified by world region after exclusion of studies with prevalence > 0.20. Regional and overall pooled estimates (common-effect and random-effects) are presented, with corresponding 95% confidence intervals and heterogeneity measures, demonstrating persistent but attenuated between-region variability. CI: Confidence interval.

Figure 6 Forest plot of chronic kidney disease in young people. A: Forest plot of chronic kidney disease (CKD) prevalence in young people stratified by world region after further restriction to studies with prevalence ≤ 0.16. Regional pooled estimates and overall pooled prevalence (common-effect and random-effects) with 95% confidence intervals are shown, illustrating that substantial heterogeneity remains despite exclusion of the highest-prevalence studies; B: Forest plot of CKD prevalence in young people (≤ 25 years) by study setting in low- and middle-income countries. Studies are grouped as community-based, hospital/clinic-based, CKD/end-stage renal disease registries, and specific disease cohorts. Squares represent individual study estimates (size proportional to study weight), horizontal lines indicate 95% confidence intervals, and diamonds show pooled prevalence for each setting and overall, obtained from random-effects generalized linear mixed models (logit transformation). CI: Confidence interval.

An a priori classification by study setting further clarified these patterns (Figure 6B). When all 19 studies were stratified into community-based, hospital/clinic-based, CKD/ESRD registry, and specific disease cohorts, a clear gradient emerged. In community-based samples (7 studies), the random-effects pooled prevalence was 3.6% (95%CI: 0.8%-15.8%), whereas in hospital/clinic-based studies (9 studies) it rose to 15.2% (95%CI: 9.8%-22.9%). The single CKD/ESRD registry study reported a prevalence of 27.9% (95%CI: 25.8%-30.3%), and the two specific disease cohorts had the highest pooled prevalence at 59.2% (95%CI: 18.2%-90.4%). The test for subgroup differences by setting under the random-effects model was significant (Q between = 18.5; df = 3; P = 0.0003), highlighting the strong influence of sampling frame and clinical context on observed prevalence.

To formally quantify how much of the between-study variability could be attributed to these factors, we fitted a series of random-effects meta-regression models on the logit-transformed prevalence. In a model including only the study setting as a moderator, the overall test for moderators was significant [Q-statistic for Moderators (QM) = 11.0; df = 3; P = 0.012]. Compared with community-based studies, hospital/clinic-based samples had a higher prevalence (β = 1.53; 95%CI: 0.06-3.00; P = 0.041), and specific disease cohorts showed a still larger increase (β = 3.64; 95%CI: 1.31-5.97; P = 0.002), while the CKD/ESRD registry category had a positive but imprecise coefficient. In this model, the between-study variance was reduced from τ² = 3.41 in the unadjusted model to τ² = 2.16, indicating that study setting alone explained approximately 36.8% of the heterogeneity (pseudo-R² approximately 37%). A meta-regression including only world region as a moderator yielded a borderline overall moderator effect (QM = 7.21; df = 3; P = 0.066), with higher prevalence in African studies relative to the reference region (β = 2.81; 95%CI: 0.63-4.99; P = 0.011) and a trend towards higher prevalence in Asia. This region-only model reduced τ² to 2.45, corresponding to approximately 28.2% of the heterogeneity statistically attributed to region.

In the fully adjusted model including both setting and region, the joint test for moderators was strongly significant (QM = 29.17; df = 6; P < 0.001). In this model, specific disease cohorts (β = 3.33; 95%CI: 1.22-5.43; P = 0.002), CKD/ESRD registries (β = 3.27; 95%CI: 0.67-5.88; P = 0.014), and African region (β = 3.16; 95%CI: 1.14-5.18; P = 0.002) remained independently associated with higher logit-transformed prevalence. The residual between-study variance fell to τ² = 1.32, implying that setting and region together accounted for approximately 61.4% of the heterogeneity observed in the unadjusted model. Despite this substantial reduction, residual heterogeneity remained considerable, consistent with ongoing variation in age structure, CKD definitions, and unmeasured context-specific factors across the included studies. Ratings for risk of bias were based on the 9-domain Joanna Briggs Institute prevalence checklist[21], and summarized graphically using the RobVis tool (Figure 7)[9,10,19,20,22-36]. The traffic-light plot shows that most studies had low risk for sampling method, sample size, and statistical analysis, but several were judged at high risk in relation to sample frame, analysis coverage, and response rate, which likely contributes to the substantial between-study heterogeneity observed in the meta-analysis.

Figure 7 Traffic-light plot of risk-of-bias assessment for included prevalence studies. Each row represents a study and each column one of the nine Joanna Briggs Institute prevalence domains (D1-D9), with green circles indicating low risk of bias, yellow unclear risk, orange high risk, and blue indicating no information. The right-hand column summarizes the overall judgement for each study.

DISCUSSION

This study, which set out to deepen the understanding of CKD in young patients living in developing countries, confirms that the burden of disease in this age group is substantial. In our primary random-effects generalized linear mixed model, the pooled prevalence of CKD among individuals ≤ 25 years in LMICs was 11.55%, with a very wide prediction interval and extreme heterogeneity, and even when analyses were restricted to more general community or mixed populations, the pooled prevalence remained in the low single-digit range, but still clinically important. Nearly all included studies reported non-trivial levels of CKD, and our systematic review and meta-analysis add to the evidence base by bringing together recent prevalence estimates and formally quantifying how geography and study setting drive the wide variability observed across LMICs. The magnitude of CKD burden observed in individual studies is consistent with previous reports. For example, Garcia-Garcia et al[9] found a CKD prevalence of 10.5% in Jalisco, Mexico, very close to the average of more general populations in our sensitivity analyses, whereas Ndongo et al[35] reported a strikingly high prevalence of 17.8% in northern Senegal. In our dataset, prevalence estimates ranged from 0.16% in a Bolivian community survey to 84.9% in a small high-risk Indian cohort, and our subgroup and meta-regression analyses showed that much of this spread reflects differences in sampling frame and underlying risk profile rather than random variation alone.

By classifying studies into community-based samples, hospital/clinic populations, CKD/ESRD registries, and disease-specific cohorts, we demonstrated a clear gradient in prevalence from the lowest levels in community surveys to the highest in registries and specific high-risk cohorts. Study setting alone explained roughly one-third of the between-study variance, while world region accounted for a further portion, with particularly high adjusted prevalence in African studies compared with other regions. When both setting and region were included in a meta-regression model, together they explained more than half of the observed heterogeneity, indicating that structural features of where and how young participants were sampled are major determinants of the reported prevalence rather than purely biological differences or chance.

Only modest rural-urban differences were reported in some studies[9], but our findings suggest that even small observed gradients may mask substantial pockets of undiagnosed disease in disadvantaged communities, particularly where screening is limited and laboratory capacity is constrained. Young people from low socioeconomic backgrounds are likely to bear a disproportionate burden of CKD in LMICs, yet race and ethnicity - highlighted as important risk factors in the global CKD literature - were not systematically examined in the included studies. This lack of granularity limits our ability to disentangle biological from social determinants and should be a priority for future research. Gautam et al[25] further illustrated this point by showing that, although urinary tract infections were the most common diagnosis, substantial CKD prevalence was also observed among pediatric patients in tertiary hospitals, underscoring how hospital-based cohorts can over-represent more severe or complicated disease. Our setting-based analyses are consistent with this pattern, with tertiary and registry samples contributing disproportionately high prevalence estimates compared with community-based surveys.

Overall, this systematic review of 19 studies underscores the urgent need for broader, methodologically consistent epidemiological research on CKD prevalence in young populations across LMICs. While the pooled estimates for more general populations clustered in the range of approximately 5%-10%, they were accompanied by very wide prediction intervals, reflecting the influence of geographical, socioeconomic, and health-system factors. These findings are concordant with data from the Philippines[29], where CKD prevalence ranged from 7.0% to 34.3%, and with estimates of 10.7% across four African countries and 11.2% across 14 Asian countries[15], further illustrating the global variability in CKD burden and the need for standardized approaches to measurement and reporting.

The marked heterogeneity observed in our analysis is likely multifactorial. In sub-Saharan Africa, for example, high rates of comorbid conditions such as HIV/acquired immunodeficiency syndrome are known to increase CKD risk and may contribute to higher prevalence estimates in some sites. In many LMICs, limited access to essential medications and services for diabetes and cardiovascular disease, together with constrained infrastructure for early detection and management of both non-communicable and communicable renal diseases, may accelerate CKD progression and delay diagnosis. These patterns are consistent with the notion that increases in gross domestic product, literacy, and universal health coverage can help reduce CKD prevalence by improving prevention, detection, and treatment capacity in developing countries[30].

Beyond study design and health-system factors, environmental exposures remain an important contextual consideration. Occupational and environmental contact with heavy metals and nephrotoxic chemicals - through mining activities, contaminated water sources, or workplace solvents - can cause or exacerbate chronic kidney damage, with clinical manifestations ranging from subtle tubular dysfunction to overt kidney failure. Well-described nephrotoxic agents include arsine gas, carbon tetrachloride, aristolochic acid, and exposures associated with Balkan endemic nephropathy and Chinese herb nephropathy[31]. Although our meta-analysis could not directly adjust for these exposures, they likely contribute to the background risk in certain regions and may partly explain the high prevalence in specific localities.

Infectious diseases such as HIV, hepatitis B, and hepatitis C are also recognised contributors to renal impairment, and region-specific infections can precipitate acute kidney injury that evolves into chronic damage. Even after apparent recovery from acute episodes, residual scarring and subclinical loss of nephron mass can increase long-term CKD risk[31]. While these mechanisms were not routinely captured in the included studies, they form part of the broader epidemiological context in which our prevalence estimates should be interpreted.

A country-level reading of our data illustrates how environmental exposures, comorbidities, and study design interact with the setting- and region-level sources of heterogeneity demonstrated in our subgroup and meta-regression analyses. In several countries, CKD prevalence in young people appears to sit on a background of substantial cardiometabolic and environmental risk. In Albania, for example, obesity, cardiovascular disease, diabetes, and stress-related conditions are recognised precipitating factors for CKD and are increasingly targeted through interventions on alcohol, smoking, depression, and diet[31]. At the same time, long-term surveys have documented very high concentrations of Pb, Ni, Cr, and Co in Albanian soils, often exceeding European Union pollution limits and showing strong correlations with soil characteristics such as clay content and cation exchange capacity[32,34]. Similar patterns of mixed cardiometabolic and environmental risk appear in high-prevalence settings in Africa and Asia, where our highest study-level estimates were observed.

In a number of countries, mining and industrial activity with documented heavy-metal contamination provides a plausible environmental backdrop to the CKD burden observed in young populations. In Bolivia, for instance, Plata et al[32] reported a very low prevalence (0.16%) in a large community survey of 14082 individuals, but this may partly reflect sampling and case-finding limitations rather than genuine protection. Independent environmental assessments from North Potosí show severe contamination of rivers used for drinking and irrigation with arsenic, cadmium, copper, iron, lead, nickel, and zinc at concentrations far above recommended limits[34], raising concern that CKD may be under-recognized rather than rare in this context. In Nigeria and South Africa, artisanal and industrial mining of gold and other metals exposes workers and surrounding communities to lead, mercury, arsenic, cadmium, and related nephrotoxins; studies in Nigerian mining communities have demonstrated markedly elevated blood metal levels and impaired kidney function in both occupationally and environmentally exposed individuals compared with controls[35]. Sediment studies from South African river systems likewise document moderate to considerable enrichment of several heavy metals, with predicted ecological and potential renal risks[14]. These observations are consistent with the high hospital- and clinic-based prevalence we observed in some African cohorts and help contextualize the extreme between-study variability.

China, India, Iran, and Guatemala illustrate further ways in which industrialization and occupational exposures may shape CKD risk in young people. In China, population-based work has shown that daily exposure to mixtures of metals and metalloids - particularly copper and arsenic - is associated with albuminuria, reduced eGFR, and higher CKD risk[37,38]. In India, coal and mineral extraction involve multiple nephrotoxic metals (arsenic, manganese, copper, lead, chromium, iron, and mercury), with additional arsenic loading of water bodies from Himalayan runoff[14,39]. At the same time, traditional medicines and nephrotoxic drugs remain relevant exposures, and there is growing interest in using more sensitive biomarkers to detect early renal injury and to evaluate potentially nephroprotective therapies[37]. In Iran, the Hovayzeh cohort found urinary concentrations of several metals above reference values in both CKD patients and controls; although associations with CKD were not statistically conclusive, the findings highlight potential health concerns and the need for larger samples[38]. In Guatemala, sugarcane workers exposed to heat stress and low-level metal exposure showed inverse associations between urinary cadmium/arsenic and eGFR, and positive associations with neutrophil gelatinase-associated lipocalin, suggesting that even sub-threshold exposures may contribute to CKD of unknown etiology in vulnerable occupational groups[34].

Other countries in our dataset show lower CKD prevalence in young populations despite documented environmental contamination, underscoring the complexity of exposure-response relationships and the limitations of the available epidemiology. Cambodia and Malawi, for example, contributed low-prevalence community or mixed samples in our meta-analysis, with relatively low risk of bias, yet both have water bodies and reservoirs affected by agricultural runoff and urban waste, with measurable heavy-metal loads[13,36,37]. In Venezuela and Saudi Arabia, CKD prevalence in young people in the included studies was also relatively low, while environmental surveys of coastal sediments (Venezuela) and urban dust (Riyadh) confirm heavy-metal contamination above background in areas of intense human activity[22,25]. In these settings, the discordance between environmental measurements and epidemiological estimates may reflect true differences in susceptibility, shorter exposure duration in younger age groups, or simply limitations of cross-sectional designs with crude kidney outcomes and incomplete case ascertainment.

Egypt and Senegal illustrate high prevalence in the context of both non-communicable and environmental risk. Egyptian studies in our review reported substantial CKD burden in children and young patients, with diabetes, nephrolithiasis, urinary tract infections, congenital anomalies, and primary glomerulonephritis among the leading causes, and a sizeable fraction of cases with unknown etiology[39,40]. Parallel geochemical work from Egyptian coastal and desert regions describes elevated levels of iron, manganese, zinc, nickel, copper, cadmium, and lead in sediments, attributable to terrigenous inputs, land-based pollution, and legacy gold mining activities[40,41], adding a plausible environmental layer to more classical CKD risk factors. In Senegal, very high CKD prevalence in the included hospital-based cohort coincides with serious soil contamination by waste electrical and electronic equipment in Dakar, with “very high risk” classifications for lead, cadmium, and mercury[35]. While our meta-analytic models could not explicitly incorporate metal exposure, these country-specific data support the hypothesis that heterogeneity partly reflects varying mixtures of infectious, cardiometabolic, occupational, and environmental drivers.

Finally, Mexico and several Latin-American sites demonstrate that even where sampling frames are closer to the general population, CKD prevalence in young people can still be appreciable. The Jalisco study reported a prevalence of around 10.5% in a large Mexican cohort[9], and national environmental reviews document widespread exceedance of permissible limits for multiple heavy metals in soil, sediment, and water across many states, driven by mining, industry, agriculture, and urban wastewater[34]. Community-based studies from Bolivia and Venezuela showed much lower prevalence in our pooled analysis[22,34], but in all three countries, the contrast between environmental contamination and divergent epidemiological estimates reinforces the need for standardized case definitions, consistent eGFR estimation in children and young, and better-designed population surveys.

Taken together, these country-level observations align with our quantitative findings that study setting (community vs hospital/clinic vs registry or disease-specific cohort) and geographical region together explain a substantial proportion of the between-study variance, while leaving considerable residual heterogeneity[31,32]. They also highlight that environmental metal exposure, occupational heat stress, comorbid infections (such as HIV), and the rising burden of non-communicable diseases likely contribute to the complex and uneven distribution of CKD in young people across LMICs, but remain under-measured in most existing prevalence studies[13,21,33-37].

The limitations in CKD research are numerous. Publication bias often favors studies with positive or significant results, potentially underestimating the true prevalence and impact of CKD. Variability in diagnostic criteria and disease definitions across studies can lead to misclassification and inaccurate estimates, making direct comparisons difficult. Additionally, many studies offer only a cross-sectional view, failing to capture the progression and long-term effects of CKD. Research from LMICs is frequently underrepresented, resulting in data gaps and possible underestimation of CKD burden in these regions. Addressing these challenges is crucial for producing more comprehensive and reliable data, which will enhance CKD management strategies and improve patient outcomes[42].

More broadly, the complexity and heterogeneity observed in our findings mirror challenges described across other disease domains in LMICs, where epidemiological signals often emerge at the intersection of health-system limitations, infectious burden, environmental exposures, and delayed implementation of standardized protocols. Prior work from our group and others has shown that, in resource-constrained settings, delayed diagnosis, fragmented care pathways, and inconsistent application of evidence-based frameworks can substantially influence disease detection, severity at presentation and outcomes, whether in acute conditions managed through telemedicine platforms, protocol-driven management of hepatorenal syndrome, or chronic infectious and inflammatory diseases prevalent in LMICs[37-41]. Similarly, systematic syntheses focusing on infections and organ-specific complications in developing regions highlight how underdiagnosis, late referral, and contextual factors - rather than biological rarity alone - often explain apparently discordant prevalence estimates[40,41]. These considerations are directly relevant to CKD epidemiology in young populations, where subclinical disease, prior infectious insults, and cumulative nephrotoxic exposures may remain unrecognized until advanced stages, reinforcing the need to interpret pooled prevalence estimates within a broader structural and methodological context.

Despite extensive subgroup and meta-regression analyses, between-study heterogeneity remained very high, indicating that unmeasured factors - such as differences in diagnostic practice, healthcare access, environmental exposures, and underlying risk profiles - likely contribute to the wide variation in CKD prevalence estimates. Study designs and sampling frames varied considerably, ranging from community surveys to high-risk clinical and registry cohorts, which may respectively under- and overestimate the true prevalence in the general young population and limit generalizability. Case definitions and methods for estimating kidney function were not fully standardized, with some studies using adult eGFR equations in pediatric or young patients, potentially misclassifying CKD status. Although we attempted to mitigate this by applying strict inclusion criteria, excluding obvious high-risk cohorts in sensitivity analyses, and modelling setting and region, residual confounding and misclassification cannot be excluded. Finally, the analysis relied on published data from LMICs, where underdiagnosis, underreporting, and publication bias are likely; as such, our pooled estimates should be interpreted as approximations of the underlying burden rather than precise measures of true prevalence.

Patients in agricultural communities face a high risk of CKD of unknown etiology, due to combined geo-environmental and agro-environmental factors, such as climate variability and exposure to agricultural chemicals. While research on how these chemicals specifically affect kidney disease progression is limited, new analytical models that link clinical and environmental data are starting to reveal the impact of agricultural practices on kidney health in rural areas[43]. Hospital-based studies focus on inpatients, helping identify undiagnosed cases and initiate treatments, while community-based studies look at CKD in the general population, highlighting risk factors and stressing early detection. Both study types complement each other as hospital-based research improves patient care within hospitals, while community-based research shapes public health policies. Together, they provide a full picture of CKD prevalence and guide effective prevention and treatment strategies[44].

Taken together, our results argue strongly for earlier and more systematic detection of CKD in young populations in resource-constrained settings. Priority actions include implementing standardized diagnostic criteria and age-appropriate GFR equations, integrating CKD screening into existing child and young patient health platforms, and establishing national or regional CKD registries to provide reliable surveillance data. Strengthening primary care, improving access to nephrology services, and addressing upstream drivers - such as uncontrolled non-communicable diseases, nephrotoxic exposures, and environmental contamination - will be essential to slow progression and reduce long-term complications. CKD in young people in LMICs is not a rare complication but an emerging public health challenge; recognizing and quantifying this burden is the first step towards designing realistic, context-appropriate strategies to prevent avoidable kidney failure over the life course.

CONCLUSION

This meta-analysis shows that CKD in young people (≤ 25 years) in LMICs is both common and highly variable. Across 19 studies from 17 countries, the overall pooled prevalence was around 12%, with even sensitivity analyses restricted to more general community and mixed populations still indicating a non-trivial burden in the range of 4%-5%. Much of the extreme heterogeneity was explained by study setting and region: Prevalence was lowest in community-based samples and highest in registries and high-risk clinical cohorts, with meta-regression suggesting that setting and geography together accounted for a substantial proportion of between-study variance. These findings indicate not only genuine differences in risk profiles but also major inconsistencies in how CKD is identified, investigated, and reported in young people across LMICs.

ACKNOWLEDGEMENTS

We extend our appreciation to the Faculty of Life Sciences and Education at the University of South Wales, in partnership with Learna Ltd., for the Gastroenterology MSc program and their invaluable support in our work. We sincerely acknowledge the efforts of the University of South Wales and commend them for their commitment to providing lifelong learning opportunities and advanced life skills to healthcare professionals.