Global and national burden of chronic kidney disease attributable to fine particles: An updated exposure-response meta-analysis Authors: Pengfei Li, Xueqiu Ni, Jingyi Wu, Chenghua Guo, Tao Xue, Luxia Zhang Published: 2026 ABSTRACT AbstractRobust evidence is needed for the exposure-response function (ERF) linking chronic kidney disease (CKD) to long-term exposure to ambient fine particulate matter (PM2.5). A three-stage mixed-effects meta-analysis framework was applied to estimate a nonlinear ERF that links PM2.5 exposure to CKD risk ratio or estimated glomerular filtration rate (eGFR) reduction, by integrating all available evidence via systematic searches. Based on the derived ERF, we quantified the global and national burden of CKD attributable to PM2.5 exposure above 10 μg/m3 from 1990 to 2020. A total of 38 studies on CKD and 16 on eGFR were included. Per 10 μg/m3 increment in PM2.5, the pooled risk ratio of CKD was 1.32 (95% confidence interval [CI]: 1.21–1.43) and the ERF indicated a predominantly log-linear association; the pooled eGFR reduction was 2.115 mL/min/1.73 m2 (95% CI: 0.452–3.777), and the ERF showed a saturated effect of high-concentration exposure. In 2020, PM2.5 exposure was estimated to contribute to 485,000 (95% CI: 370,000–587,000) deaths, 5.77 million (95% CI: 4.51–6.99 million) incident cases, and 243 million (95% CI: 188–294 million) prevalent cases of CKD worldwide. Compared with level-2 risk factors for CKD mortality in the Global Burden of Disease study, PM2.5 ranked first in 1990 and second in 2020, reaching a level of importance comparable to major metabolic risk factors. This study provides robust evidence that PM2.5 is a substantial contributor to the global CKD burden and that clean air action is an urgent and integral CKD prevention.Keywords: fine particulate matter, chronic kidney disease, exposure-response function, risk assessment Introduction Chronic kidney disease (CKD) is a major contributor to the global diseases burden, leading to premature mortality, increased risk of other non-communicable diseases (NCDs), and decreased quality of life.1 Based on the estimates from the global burden of diseases (GBD) assessment, over 850 million people worldwide live with kidney disease, with a global median prevalence of 9.5% (interquartile range [IQR]: 5.9–11.7).2,3 In recognition of this burden, the 78th World Health Assembly in 2025 endorsed a resolution calling for strengthened prevention, control, and promotion of kidney health within the global noncommunicable diseases agenda.4 Notably, the burden is unevenly distributed, with low- and middle-income countries (LMICs) disproportionately and increasingly affected.3 While diabetes, hypertension, and obesity are well-documented risk factors of CKD, the role of environmental risk factors, particularly ambient air pollution, has gained increasing attention.5,6 Disadvantageous air quality might be a key driver of the concentrated spatial pattern and growing trend of CKD in LMICs. Fine particulate matter (PM2.5), airborne particles with a diameter of 2.5 μm or less, can penetrate deep into the lungs, enter the bloodstream, and induce systemic inflammation, oxidative stress, and endothelial dysfunction.7 These pathological pathways are known to contribute to kidney damage. Previous studies have largely focused on PM2.5-attributable burden of ischemic heart disease, stroke, chronic obstructive pulmonary disease, lung cancer, acute lower respiratory infection, and type 2 diabetes mellitus, as established by GBD framework.8,9 More recently, emerging epidemiological evidence has expanded this scope by linking long-term PM2.5 exposure to increased risk of CKD incidence and prevalence, as well as accelerated progression to end-stage kidney disease (ESKD).5,6,10 Particularly, due to the nature of prolonged and universal exposure, the contribution of PM2.5 to the global disease burden at the population level is substantial, although its effect on an individual may be modest. Therefore, PM2.5-associated CKD is a problem of significant public health importance but is easily neglected, especially when compared to traditional clinically focused risk factors such as diabetes. Additionally, quantitative assessments to evaluate the burden of PM2.5-associated CKD are scarce, because the exposure-response relationship is insufficiently studied. A landmark study by Bowe et al. was pivotal in synthesizing evidence to characterize a non-linear exposure-response function between PM2.5 and CKD.6 For the first time, they utilized the tailored function to evaluate the global burden of PM2.5-associated CKD. Most importantly, they showed that the relationship was sublinear, with the rate of risk increase attenuating at higher PM2.5 concentrations. Their work also revealed that models relying solely on ambient PM2.5 data estimated a greater risk than integrated models that included proxy exposures (e.g., smoking). However, their approach drew from the Global Exposure Mortality Model (GEMM), which pre-supposed a curvature to estimate the exposure-response function. This assumption may lead to an underestimated burden of PM2.5-associated CKD at high-concentration regions, which are highly overlapped with the data-poor LMICs. Furthermore, about six years after the landmark study, there are more and more epidemiological studies to explore the relationship between PM2.5 exposure and CKD,11,12,13 which makes it possible to develop an estimate purely based on ambient PM2.5 (without the inclusion of proxy exposures). To fill the knowledge gap, our study introduces a novel nonlinear exposure-response function (ERF) between PM2.5 and CKD. Compared to the previous work, we focus exclusively on ambient PM2.5 exposure, aiming to provide a more specific and targeted risk assessment. Furthermore, we introduce a novel three-stage method to develop the ERF that does not pre-suppose a sublinear relationship. The method allows for a data-driven curvature, which may generate more accurate burden estimation, particularly in high-exposure regions. Finally, we apply it to provide the updated estimates of the global and national CKD burden attributable to PM2.5 from 1990 to 2020. Materials and methods Literature review and meta-analysis To characterize the nonlinear exposure-response relationship between PM2.5 and CKD across the full range of human exposures, we conducted an integrative meta-regression of published evidence. Following PRISMA guidelines, we systematically searched PubMed, OVID Embase, and Web of Science for cohort, case-control, and cross-sectional studies examining the association between PM2.5 and CKD, as well as studies linking PM2.5 with estimated glomerular filtration rate (eGFR), a key indicator of kidney function. Detailed search strategies are provided in Table S1. The search, conducted on July 18, 2024, identified 7,896 potentially relevant studies. Studies were screened according to the following criteria: published in a peer-reviewed journal; full text available in English; cohort, case-control, or cross-sectional design; assessment of kidney-related outcomes; evaluation of long-term PM2.5 exposure; and provision of relative risk (RR) estimates or relevant estimates that can be converted to RR, such as odds ratio (OR), hazard ratio, and excess risk. The primary outcome was CKD-related outcome (including incidence, prevalence, and mortality) following established practice in GBD assessments9,14 and previous integrated ERF modeling frameworks,6 and the secondary outcome was change in eGFR (mL/min/1.73 m2). From eligible studies, we extracted information on study design, participants, outcomes, exposure levels, relative risks, uncertainty estimates, model settings, and relevant factors for risk-of-bias assessment. After applying these criteria, 39 studies focusing on PM2.5 and CKD10,11,12,13,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49 and 16 on PM2.5 and change in kidney function13,15,16,17,18,19,20,21,22,23,24,50,51,52,53,54 were included in our analyses. The PRISMA flowchart is provided in Figure 1 and additional details of included studies are provided in Table S2. PRISMA flowchart We evaluated study quality and risk of bias using three complementary tools: the Office of Health Assessment and Translation (OHAT) Risk of Bias (RoB) tool,55 the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system,56 and the Newcastle-Ottawa Scale (NOS) approach.6,57 Each study was independently reviewed by two team members, with disagreements resolved by a third. The OHAT tool assessed the internal validity of studies, focusing on whether design or conduct compromised the credibility of the reported associations. According to OHAT criteria (Table S3), none of the included studies were rated as “definitely or probably high” risk of bias. The GRADE system was used to evaluate the overall certainty of evidence for each outcome, considering potential downgrades for bias, imprecision, inconsistency, indirectness, or publication bias. In our analysis, the certainty of evidence for both CKD and eGFR outcomes was rated as moderate (Table S4). Finally, the NOS tool assessed non-randomized studies based on group selection, comparability, outcome or exposure assessment, and adjustment for importing confounders. All included studies scored above 50% of the maximum NOS score (Table S5), and the confounders for the main adjustment of the studies were provided in Table S6. Together, these evaluations strengthened confidence in the robustness and reliability of our meta-analysis findings. We selected the OHAT framework as our primary RoB tool because it is a general-purpose instrument applicable to diverse environmental exposures and provides detailed criteria for evaluating bias. It has been widely applied in environmental epidemiology systematic reviews.58,59 Moreover, OHAT ratings can be directly linked to the GRADE approach for assessing certainty of evidence, facilitating comparison of evidence strength across reviews that employ different RoB tools. Although the core bias domains addressed by OHAT are largely consistent with those in the World Health Organization (WHO)-recommended RoB instrument, direct comparability of absolute risk-of-bias ratings with reviews using the WHO tool may be limited. We conducted meta-regression analyses to estimate a pooled effect of PM2.5 on CKD using 38 effect estimates out of the 39 CKD studies (excluding one that examined ESKD survival rate49) and a pooled effect for change in eGFR using 16 studies. The publication bias tests using funnel plots were provided in Figure S1. For sensitivity analyses, we first varied outcome definitions and estimated pooled effects for CKD incidence, CKD prevalence, ESKD incidence, CKD mortality, change in eGFR based on the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation,60 and change in eGFR based on the Modification of Diet in Renal Disease (MDRD) Study equation.61 Second, sub-group meta-regression analyses were also performed to assess pooled effects of PM2.5 across different subgroups. Third, we applied a leave-one-out approach—a widely used machine learning technique for detecting potential outliers—and re-ran the meta-regression analyses after trimming (Figure S7). Three-stage model for nonlinear exposure-response function To provide a more specific assessment of the ERF between PM2.5 and CKD, denoted as f(X), we developed a three-stage approach that integrates all relevant evidence without assuming a sublinear relationship. In the first stage, we characterized study-specific PM2.5 exposures. For each included study, we estimated the distribution of PM2.5 concentrations at the study location and derived a calibrated mean value for the study population. We assumed exposures followed a gamma distribution, X ∼ Gamma(k, θ), where X is PM2.5 concentration, k is the shape parameter, and θ is the scale parameter. Ambient PM2.5 concentrations are strictly non-negative and typically right-skewed; the gamma distribution offers a flexible parametric form capable of capturing such skewness and heterogeneity and has been widely used in air pollution exposure modeling.62,63 The gamma distribution parameters were estimated using all available summary statistics of PM2.5 exposures reported in each study (i.e., mean, standard deviation, median, and quartiles) with maximum likelihood estimation method. The calibrated mean PM2.5 exposure for each study was then derived from the fitted gamma distribution. In the second stage, we applied a mixed-effects meta-analysis model to estimate the nonlinear marginal effect of PM2.5 on CKD.64 Specifically, we regressed log-relative risks, log(RR), on the calibrated mean PM2.5 concentrations, incorporating both fixed and random effects across studies. The resulting marginal effect estimates describe how small increases in PM2.5 exposure alter CKD risk, representing weighted averages of the first-ordered derivative of the ERF between PM2.5 and CKD under certain assumptions, denoted as f'(X), according to Yitzhaki’s theorem and its extension.65 The mixed-effects model can be parameterized as follows: where RR is the marginal relative risk of CKD per 10 μg/m3 increment in PM2.5 at the exposure value X, ns(X, df = 3) denotes a nonlinear natural spline function of PM2.5 exposure with 3 degrees of freedom, Zu is the study-level random effects, and e is the error term. The mixed-effects model was fitted via maximum likelihood estimation method using the rma.mv function in R. In the third stage, we applied an integration of f'(X) given a predetermined theorical minimum risk exposure level (TMREL) to obtain the nonlinear ERF between PM2.5 and CKD, denoted as f(X), using the following equation: where log(RRTMREL) denotes the log-transformed risk ratio of CKD referring to TMREL, derived from the fitted first-ordered derivative curve of ERF between PM2.5 and CKD. Considering the empirical exposure distribution of the included studies in the low-concentration PM2.5 range (Table S2), the extrapolation estimates in this range are subject to high uncertainty. Therefore, TMREL is set to the fourth-stage interim target (10 μg/m3) based on the Air Quality Guideline (AQG) from the WHO for annual PM2.5 exposure. Using the three-stage method, we developed nonlinear ERFs linking PM2.5 with CKD-related outcomes and with change in eGFR, incorporating all available evidence. Our approach interprets study-specific effect estimates as local marginal effects and synthesizes them to reconstruct a continuous exposure-response function. This framework is consistent with well-established methodologies used to derive air-pollution exposure-response functions, including the integrated exposure-response model,14 GEMM,66 and meta-regression-Bayesian, regularized, trimmed (MR-BRT) adopted in GBD assessments.9,67 For sensitivity analyses, we derived a nonlinear ERF for CKD restricted to incidence-based studies, and an ERF for change in eGFR limited to studies that applied the CKD-EPI equation. Disease burden assessment Utilizing the derived nonlinear ERFs between CKD-related outcomes and PM2.5, we estimated the global and national burden of CKD attributable to PM2.5 from 1990 to 2020. Country-level data on CKD incidence, prevalence, and mortality were obtained from the 2021 GBD study.3 These were linked with annual global estimates of population-weighted PM2.5 from 1990 to 2020 provided by GBD investigators (Table S9).8 The national burdens of CKD attributable to PM2.5 were calculated using the following equations: where s denotes the index of country or territory, t is the index of calendar year, h is a specific CKD-related outcome (incidence, prevalence, or mortality), f(Xs,t) denotes the estimated risk ratio of CKD referring to TMREL based on the derived nonlinear ERF between PM2.5 and CKD, P is the population number, N is the number of cases for a specific CKD-related outcome, AF is the attributable fraction, AN is the attributable number, and AR is the attributable number per million population. Country-level population data were obtained from the 2021 GBD study.68 We used a Monte Carlo approach to generate 95% confidence intervals for disease burden estimates, incorporating uncertainty from both risk estimation and the GBD burden data. To investigate the spatiotemporal trend of PM2.5-associated burden of CKD, we further calculated the annual relative change of CKD deaths attributable to PM2.5 by countries using the Theil-Sen slope estimator. Additionally, we conducted a comparative assessment of our estimates on PM2.5-attributable CKD mortality with the burden associated with level 2 risk factors reported in the GBD 2021 study. All data analyses were conducted using R software (v.4.1.3; The R Foundation for Statistical Computing, Vienna, Austria). The traditional meta-regression analyses were implemented using the R package meta, and the mixed-effects meta-analysis model was conducted with the metafor package. Results Literature review and meta-analysis Our systematic search identified 39 studies focusing on CKD and 16 on change in kidney function eGFR, meeting the inclusion criteria. The geographic distribution of these studies is shown in Figure 2A. The studies were concentrated in North America, Europe, and East Asia, particularly in China (22 studies) and the USA (7 studies). Study locations were in regions with exposure levels varying from low to high PM2.5 concentrations. The included effect estimates are appropriate to develop an exposure-response function due to the wide range of support. The classical meta-analysis for CKD risk, encompassing 38 effect estimates out of the 39 studies (excluding one study that looked at the ESKD survival rate), yielded a pooled risk ratio of 1.32 (95% confidence interval [CI]: 1.21–1.43) per 10 μg/m3 increment in PM2.5 (Figure 2B). For specific outcomes, the pooled risk ratios were estimated as 1.46 (95% CI: 1.29–1.65), 1.14 (95% CI: 1.07–1.22), 1.62 (95% CI: 1.13–2.32), and 1.19 (95% CI: 1.04–1.37) for CKD incidence, CKD prevalence, CKD mortality, and ESKD incidence, respectively (Figure S2). The meta-analysis for change in kidney function, based on 16 studies, showed a pooled mean decrease in eGFR of 2.115 mL/min/1.73 m2 (95% CI: 0.452–3.777) per 10 μg/m3 increment in PM2.5 (Figure 2C). Different methods to evaluate eGFR didn’t significantly change the pooled effect estimate (p value = 0.90) (Figure S3). For the commonly used CKD-EPI equation, the subset meta-analysis generated a similar point estimate (1.995) but with a wider uncertainty range (95% CI: −0.1691–4.159). Literature review and meta-analysis (A) Geographic distribution of the included studies. Background color shows the average PM2.5 concentration from 2001 to 2020. Black dots are used to show the studies that reported city-level locations. The studied countries are shown with a code that combines the two-digit ISO and number of individual studies conducted in that country. (B and C) Meta-analyses for the risk ratio of chronic kidney disease (CKD) and change in estimated glomerular filtration rate (eGFR), respectively. Error bars indicate 95% confidence intervals. We observed significant between-study heterogeneity in the meta-analyses for CKD risk and change in eGFR (I2 > 95%), which could be attributed to the differences in study design, study population, and exposure assessment. The pooled effect estimates for CKD and change in eGFR across different study designs and subgroups are summarized in Table S7. Specifically, we found a higher pooled risk ratio in studies conducted in Europe and Asia. Studies using 2-year annual average exposures of PM2.5 also reported a slightly higher pooled risk ratio. However, these subgroup findings should be interpreted with caution due to the limited number of studies available within several specific categories. Non-linear ERFs The newly developed ERFs demonstrated clearly nonlinear relationship between long-term PM2.5 exposure and kidney-related outcomes (Figure 3; Table S8). Figure 3 presents the integrated ERFs showing the risk ratio of CKD referring to a predetermined TMREL (10 μg/m3) at different exposure levels, and Figure S4 presents their first-ordered derivative functions (red lines) showing the estimated marginal effects of CKD per 10 μg/m3 increment in PM2.5 at different exposure values. For CKD risk (Figure S4), the marginal effect estimates reported by different studies were similar to each other, suggesting first-ordered derivative of the ERF as a constant. More specifically, the meta-regression showed the marginal effect was slightly higher at lower concentrations, and then flattened out at concentrations above approximately 50 μg/m3. Accordingly, an increase from 10 to 25 μg/m3 was associated with a slightly larger relative increase in risk than an increase from 50 to 65 μg/m3. Integration of the almost constant first-ordered derivative function (red line in Figure S4) resulted in a mostly log-linear ERF between long-term PM2.5 exposure and CKD risk (Figure 3). The sensitivity analysis that includes incidence-based studies only showed a slightly steeper ERF, which remained within the uncertainty range corresponding to our major estimator. Exposure-response function linking long-term exposure to PM2.5 to risk ratio of chronic kidney disease (CKD) Colored dots display the effect estimates reported by previous studies, and the corresponding rectangles are the 95%-percentile ranges (x axis, PM2.5 distribution; y axis, effect estimate). The bottom peaks show gamma distributions that are reconstructed from the reported summary statistics and sample size. The black solid lines and ribbons are corresponding integrations that describe the relationship between different exposure levels and different risk levels, namely the exposure-response functions. The dashed lines show the sensitivity analyses that incorporate only studies based on CKD incidence. The ERF for eGFR change showed a clearly non-linear pattern with a saturated effect at high concentrations (Figure 4). The most significant drop in eGFR occurred as PM2.5 levels rose from low to moderate concentrations (∼25 μg/m3), with the rate of decline slowing at higher exposure levels. For high-concentration exposures (>35 μg/m3), the marginal effects were estimated as statistically non-significant (Figure S4). Similarly, the sensitivity analysis using studies that employed the CKD-EPI equation for eGFR calculation only yielded similar estimates compared to the main model which further incorporated studies using the MDRD equation. Exposure-response function linking long-term exposure to PM2.5 to estimated glomerular filtration rate (eGFR) change Colored dots display the effect estimates reported by previous studies, and the corresponding rectangles are the 95%-percentile ranges (x axis, PM2.5 distribution; y axis, effect estimate). The bottom peaks show gamma distributions that are reconstructed from the reported summary statistics and sample size. The black solid lines and ribbons are corresponding integrations that describe the relationship between different exposure levels and different risk levels, namely the exposure-response functions. The dashed lines show the sensitivity analyses that incorporate only studies based on CKD-EPI equation. Global burden of PM2.5-associated CKD In 2020, we estimated that 485,000 (95% CI: 370,000–587,000) deaths, 5.77 million (95% CI: 4.51–6.99) incident cases, and 243 million (95% CI: 188–294) prevalent cases of CKD were attributable to PM2.5 exposure above 10 μg/m3 globally. In average, the exposure was estimated to reduce the eGFR by 5.9 mL/min/1.73 m2 (95% CI: 3.4–8.3). Spatial distributions of the attributable mortality burden in 1990 and 2020 are shown in Figure 5A and Tables S10 and S11. The highest risk of PM2.5-assocaiated CKD (per million) was consistently concentrated in North Africa and the Middle East (e.g., Egypt and Saudi Arabia) and parts of South America (e.g., Peru). While some regions saw a decrease in rates, many high-burden areas in 1990 remained hotspots in 2020. Temporal trends from 1990 to 2020 showed a clear increase in the absolute burden of PM2.5-attributable CKD globally (Figure 5B). The attributable mortality was nearly doubled since 1990 (i.e., 250,000 [95% CI: 187,000–308,000]), and the incidence and prevalence showed similar steep upward trends over the 30-year period. Throughout this period, Asia consistently accounted for the largest share of the global attributable burden across all three metrics. For sensitivity analysis, we first estimate global burden of CKD attributable to PM2.5 exposure above 5 μg/m3 (the WHO AQG level) and obtained 630,000 (95% CI: 484,000–749,000) deaths, 7.68 million (95% CI: 5.89–9.16 million) incident cases, and 304 million (95% CI: 236–358 million) prevalent cases of CKD attributable to PM2.5 exposure globally in 2020 (Figure S5). As expected, adopting the lower TMREL produced higher burden estimates accompanied by wider uncertainty intervals. Second, we conducted a health impact assessment based on an ERF derived exclusively from incidence-based studies and observed a slightly higher global burden of PM2.5-associated CKD (Figure S6), which nevertheless remained within the uncertainty range of our primary estimates. Attributable global burden (A) Spatial distribution of mortality caused by PM2.5-associated chronic kidney diseases (CKD) in 1990 and 2020. The colors show different levels of attributable deaths per million, and their transparencies show the corresponding degrees of statistical significance. (B) Temporal trends from 1990 to 2020 for the PM2.5-attributed mortality, incidence, and prevalence. The dashed lines show the 95% confidence intervals for the overall burden estimates. Spatiotemporal trends Analysis of national trends revealed significant heterogeneity in the change of PM2.5-associated CKD deaths between 1990 and 2020 (Figure 6A). Many countries in Southeast Asia and sub-Saharan Africa experienced a substantial annual relative increase in attributable deaths. In contrast, significantly reductions were found in North America, Oceania, and most European countries. Figure 6B shows the national distribution of attributable deaths in 1990 and 2020 and illustrates that the increasing trend in the global mortality burden can be explained by both population growth and increment in per-capita risk of PM2.5-associated CKD. Specially, Figure 6B also shows that countries like India and China had a large total burden due to their large populations, whereas countries like Saudi Arabia and Egypt had extremely high per-capita risk, which was determined by the peaked value of either baseline rate or exposure level. Crucially, when compared with other level-2 risk factors for CKD mortality from the GBD study, PM2.5 ranked as the most significant contributor in 1990 and then became the second most in 2020 (Figure 7; Table S12). Its contribution to attributable number of CKD mortality was consistently greater than that of high cholesterol, smoking, and dietary risks, and second only to high systolic blood pressure since ∼2010. Trends in global and national burden (A) Annual relative change of deaths caused by PM2.5-associated chronic kidney diseases (CKD) by countries, from 1990 to 2020. (B) Distributions of PM2.5-associated CKD deaths for each country in 2020 by rectangles, where horizontal edges are proportional to population size and vertical edges are per-capita deaths of PM2.5-associated CKD. The dashed line shows the distributions in 1990. Comparison of trends in global attributed burden of CKD between PM2.5 and level 2 risk factors identified by the global burden of diseases study (A) Contributions to the number of CKD-caused mortality. (B) Similar results for the fraction of CKD-caused mortality. Discussion In this comprehensive global analysis, we established a novel exposure-response relationship between long-term PM2.5 exposure and CKD and found a substantial and growing burden of disease attributable to this environmental risk factor from 1990 to 2020. Our findings build upon and advance the conclusions of previous landmark studies, indicating that PM2.5 is a leading, and likely previously underestimated, contributor to the global CKD burden. Our work confirms the non-linear nature of the PM2.5-CKD relationship, a conclusion also reached by Bowe et al.6 However, our study introduces critical methodological refinements. First, by focusing exclusively on ambient PM2.5, we provide a more specific risk estimate compared to models that integrated proxy exposures, which Bowe et al. noted resulted in lower risk estimates. Second, while that landmark study was limited to six studies mostly from the United States, our analysis—conducted six years later—integrated evidence from 38 studies across 11 countries spanning Europe, the Americas, and Asia, thereby providing more robust estimates of the ERF and attributable disease burden. Finally, and more importantly, our log-linear model does not assume a sublinear shape. The GEMM-based approach used by Bowe et al. showed a risk curve that flattened considerably at high concentrations. In contrast, our model, while also non-linear, may better capture the persistent risk increase at the very high PM2.5 concentrations experienced in many developing nations. This methodological difference likely explains why our burden estimates are substantially higher. For instance, our estimate of 498,000 (95% CI: 391,000–592,000) attributable deaths in 2017 surpasses the 211,000 (95% CI: 184,000–237,000) deaths estimated by Bowe et al. While some of this difference is due to the different inputs, the more robust estimation of risk at high exposure levels by this study is a primary driver. Our estimate that over half a million deaths in 2020 were attributable to PM2.5-associated CKD highlights the profound public health implications of air pollution. The continuous rise in this burden since 1990 underscores an escalating public health crisis. Notably, our analysis positions air pollution, an environmental exposure, at a level of importance comparable to major metabolic risk factors, which, however, remains absent from the current GBD risk factor framework.69 Despite recent prevention efforts targeting metabolic risk factors, progress in CKD control has been limited,69 and pollution is expected to further exacerbate the challenge. The WHO now officially recognizes air pollution as a major risk factor for NCDs,70 reinforcing the growing consensus that addressing air pollution is critical to curbing their global rise. PM2.5 poses a pervasive risk for CKD, and its reduction would yield broad health benefits. Moreover, because PM2.5 also contributes to diabetes and hypertension,71,72 two leading drivers of CKD, stringent clean air policies could mitigate the CKD burden through both direct and indirect pathways. Our findings underscore the urgency of embedding air quality management into CKD prevention frameworks, positioning clean air policies as a cornerstone of kidney health promotion. Such measures are particularly critical to counteracting the compounded effects of population aging, itself a universal risk factor for CKD. Importantly, environmental prevention strategies can also advance health equity, as disadvantaged populations often bear a disproportionate burden of environmental kidney disease.73 Our analysis revealed substantial geographical disparities in PM2.5-attributed CKD burden, with LMICs in North Africa, the Middle East, and Asia experiencing the greatest impact. These regions face a double challenge of persistently high pollution levels74,75 and health systems insufficiently equipped to address the growing CKD epidemic.76 While previous studies have attributed the disproportionate CKD burden in LMICs to demographic transitions and lifestyle changes associated with rapid urbanization,1 our findings extend this understanding by underscoring the critical role of severe air pollution. This has important implications for CKD prevention in resource-limited settings.73,77 For example, embedding air pollution control measures into urban planning and transportation policies may offer a cost-effective, population-wide strategy to mitigate the CKD burden in highly polluted regions. At the same time, innovative CKD programs should adopt region-specific approaches, prioritizing air pollution control where exposure is greatest while focusing on lifestyle-related NCD interventions in less polluted settings. This study has several limitations. First, this review was not prospectively registered in a public database; however, all methods and inclusion criteria were predefined and reported in accordance with PRISMA guidelines to ensure transparency and reproducibility. For future updates, we plan to prospectively register the review protocol and provide a completed PRISMA checklist to further strengthen methodological transparency and reporting quality. Second, our literature search was restricted to publications up to July 2024; thus, more recent studies were not included, and the accuracy of our estimates was inherently constrained by data availability. Additionally, two included studies10,26 involved overlapping populations. In conventional meta-analyses, overlapping data sources are often handled by selecting a single study based on criteria such as the most recent study period, larger sample size, more robust design, or latest publication year. However, our primary objective was to characterize the nonlinear ERF between PM2.5 and CKD rather than to estimate a single pooled linear effect. Under a nonlinear ERF framework, overlapping cohorts may still contribute nonredundant information when they span different exposure ranges. Therefore, we retained both studies in the primary analysis, although some degree of double-counting cannot be entirely excluded. Sensitivity analyses omitting either study produced ERFs that were highly similar to each other and the primary estimates, suggesting minimal impact on the overall conclusions. Moreover, most included studies were conducted in upper-middle-income and high-income countries and our meta-analysis was limited to English-language publications, which may limit the representativeness of the derived ERF for LMICs and non-English-speaking countries. Given prior evidence that populations in LMICs may be more vulnerable to air pollution-related health risks,78 this bias could have led to an underestimation of the CKD burden attributable to PM2.5 in LMICs. Third, CKD definitions varied across studies, with some relying on laboratory data and others on diagnostic codes, which may have introduced outcome misclassification bias. While definitions based on diagnostic codes may reflect clinical practice more directly, the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines recommended reinforcing the definition of CKD by incorporating laboratory markers of kidney damage. Fourth, we assumed that PM2.5 exposures within each included study followed a gamma distribution. If exposure levels are exceptionally high in certain regions and the empirical distribution deviates substantially from right skewness, for example, exhibiting left skewness or strong truncation at higher concentrations, this assumption may be violated. Such misspecification could bias the estimation of calibrated mean PM2.5 exposure and, in turn, lead to underestimation of the ERF at higher concentration ranges. Fifth, substantial heterogeneity was observed in the pooled analyses. Although we applied random-effects models to explore potential sources, heterogeneity was not fully explained. Differences in population demographics, exposure assessment methods, particle composition and toxicity, covariate adjustment strategies, and temporal or geographical contexts may have contributed and warrant further investigation. For example, the time frames used to define PM2.5 exposure differed across included studies. Most studies characterized exposure using annual averages (21 of 38 CKD studies), while others applied multi-year moving averages ranging from 2 to 7 years. Such heterogeneity may affect the comparability of estimates and introduce uncertainty into pooled results. Nevertheless, sensitivity analysis restricted to studies using annual average PM2.5 exposure yielded a pooled risk ratio of 1.29 (95% CI: 1.17–1.44) per 10 μg/m3 increase in PM2.5, which was broadly consistent with the main estimate. These findings suggest that the results are robust to differences in exposure averaging periods, although some residual heterogeneity cannot be fully excluded. Additionally, although different methods to evaluate eGFR didn’t significantly change the pooled point estimate (p value = 0.90) (Figure S3), the subset meta-analysis restricted to studies using the CKD-EPI equation yielded a wider confidence interval that included the null. This likely reflects reduced statistical power due to the smaller number of eligible studies and greater between-study heterogeneity within the CKD-EPI subgroup (larger τ2). Sixth, OR estimates were treated as approximations of risk ratios and were included in the pooled analysis. Although this approach may lead to slight overestimation, the global incidence of CKD is low, and ORs closely approximate RRs under this condition; thus, any resulting bias is expected to be minimal. Seventh, funnel plots indicated possible publication bias. However, sensitivity analyses using the leave-one-out approach (Figure S7) suggested that outlier studies had only a limited impact on the overall estimates. Finally, uncertainties inherent in GBD-derived estimates of both PM2.5 exposure and CKD burden may have affected the precision of our findings. Reliance on national-level GBD data for disease burden and exposure may introduce ecological fallacy, limiting the ability to draw subnational- and individual-level inferences. Within-country variation in both exposure levels and baseline CKD risk can influence attributable burden. Under an approximately log-linear ERF between PM2.5 and CKD, replacing the full population exposure distribution with a national mean is expected to introduce only modest bias, because attributable fractions tend to scale proportionally with exposure. Nonetheless, national averages may underrepresent highly exposed urban or industrial hotspots, where a disproportionate share of CKD cases and deaths may occur, leading to underestimation of national disease burden. Furthermore, if higher exposures coincide with greater baseline vulnerability (such as older age structures, higher comorbidity burden, or lower socioeconomic status), failure to account for this spatial correlation may further bias burden estimate downward and obscure health inequalities. Accordingly, our burden estimates should be interpreted as conservative, country-level averages. Future work that integrates subnational exposure surfaces with geographically resolved CKD surveillance data will help refine burden estimates and may alter absolute burdens or country rankings. In conclusion, this study provides compelling evidence that PM2.5 is a major, non-linear risk factor for CKD. By refining the ERF, we demonstrate that the attributable burden is substantial and has increased significantly since 1990. These findings underscore air pollution as a critical environmental determinant of kidney health, highlighting the urgent need to integrate air pollution control into global CKD prevention strategies. Resource availability Materials availability This study did not generate new unique materials/reagents. Data and code availability The datasets used in this study are publicly available. All analyses were conducted using R (v.4.1.3; The R Foundation for Statistical Computing, Vienna, Austria). The R codes for the data collection and epidemiological analyses, and any additional information required for reanalyzing the data reported in this paper are available from the corresponding author on reasonable request. Funding and Acknowledgments This work was supported by the National Natural Science Foundation of China (72474010, 72125009), Hangzhou Natural Science Foundation (2025SZRJJ1809), and Science and Technology Projects of Xizang Autonomous Region, China (XZ202501ZY0072). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions P.L., conceptualization, formal analysis, writing – original draft, writing – review and editing, supervision, funding acquisition; X.N., literature review, meta-analysis, formal analysis, writing – review and editing; J.W., literature review, formal analysis, writing – original draft, writing – review and editing; C.G., literature review, writing – review and editing; T.X., conceptualization, data curation, writing – original draft, writing – review and editing, supervision; L.Z., conceptualization, supervision, writing – review and editing. All authors contributed to the manuscript and approved the final version. Declaration of interests The authors declare no competing interests. Published Online: May 21, 2026 Footnotes It can be found online at https://doi.org/10.1016/j.xinn.2026.101429. Contributor Information Pengfei Li, Email: pfli@bjmu.edu.cn. Tao Xue, Email: txue@hsc.pku.edu.cn. Luxia Zhang, Email: zhanglx@bjmu.edu.cn. Supplemental information Document S1. Figures S1–S7 and Tables S1–S12 Document S2. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Figures S1–S7 and Tables S1–S12 Document S2. Article plus supplemental information Data Availability Statement The datasets used in this study are publicly available. All analyses were conducted using R (v.4.1.3; The R Foundation for Statistical Computing, Vienna, Austria). The R codes for the data collection and epidemiological analyses, and any additional information required for reanalyzing the data reported in this paper are available from the corresponding author on reasonable request.