Chronic kidney disease (CKD) is a highly prevalent chronic illness that affects an estimated 10% of the global population, with a significantly higher incidence among older people (1). CKD has been associated with an increased risk of cardiovascular events, hospitalizations, and mortality, especially in patients with kidney failure requiring dialysis (2). Mortality attributed to CKD is steadily increasing worldwide, with an estimated increase of over 30% since 2005, according to the Global Burden of Disease (3). These deaths are primarily of cardiovascular and infectious origin, accounting for up to 70% of the causes of mortality within the dialysis patient population (4).

Recently, air pollution has emerged as a potential contributor to the mortality associated with CKD (5). Air pollution is a complex mixture of gasses and particles of various sizes and compositions, originating from a range of sources such as road traffic, urban heating, industry, or agriculture (6). Since the first description of diesel inhalation-induced kidney damage in mouse models (7), the association between air pollution and kidney health has become a topic of increasing interest, with mounting evidence linking exposure to air pollutants with the incidence of CKD and kidney failure (8, 9). However, there is a paucity of data on the potential impact of air pollutants on mortality in dialysis patients, who are a particularly vulnerable population due to their advanced age and the burden of their comorbidity (10). The few published studies exhibit significant heterogeneity in their design and exposure assessment and are predominantly focusing on all-cause mortality. Moreover, conventional analyses of air pollutants on an individual basis, exclusively reported in the literature on the subject, may not fully capture the complex exposure patterns that occur in real-life settings (11, 12). Composite spatial indices considering multi-contamination have recently been developed and are now recognized as relevant indicators for the global assessment of air quality (13–15).

Based on a nationwide registry and spatial composite air pollution scoring methods, the study’s primary objective is to evaluate the potential impact of multiple exposures to air pollutants on mortality in incident dialysis patients. Secondary objectives include examining the association between multiple exposures to air pollutants and specific causes of mortality, such as cardiovascular and infectious-related deaths.

During the 2009–2020 period, the median weighted levels and ranges [min - max] for PM_2.5, PM₁₀, and NO₂ in France were 11.9 [9.0–16.7] μg/m³, 17.7 [14.2–25.9] μg/m³, and 13.0 [8.6–36.0] μg/m³, respectively (Supplementary Figure S1, Supplementary Table S3). A strong correlation was observed among these air pollutants over the study period, estimated at 0.83, 0.76, and 0.66 for PM_2.5-PM₁₀, PM_2.5-NO₂, and PM₁₀-NO₂, respectively. The first principal component of the PCA, based on the values of the three air pollutants, explained an average of 83% of the variance (annual ranges from 69 to 84%, Supplementary Figure S2). The contributions of PM_2.5, PM₁₀, and NO₂ to the air pollution PCA score were estimated at 36.9, 34.7, and 28.4%, respectively. Overall, a notable spatial heterogeneity in the distribution of the averaged air pollution scores over the study period was observed, with higher exposure zones located in proximity to major cities and roadways (Figure 1).

The cohort included a total of 90,373 incident dialysis patients between 2012 and 2020. This was a predominantly older adult population with a mean age of 69 years, primarily composed of male subjects (65%), and approximately half of the population suffering from diabetes (45%). The vast majority of the population (89%) received hemodialysis treatment, with over a quarter initiating treatment in an emergency setting (28%). The characteristics of the patients were generally comparable according to the different levels of the air pollution PCA score, except for the socio-environmental characteristics: the patients most exposed to air pollution were most likely living in municipalities with a higher population density, less socially deprived, and closer to the nearest dialysis center (Table 1).

Over a median follow-up period of 27 months, a total of 44,242 deaths were observed, of which 8,859 were attributed to cardiovascular causes (20.0%) and 6,860 to infectious causes (15.5%). After adjusting for baseline characteristics, we observed a progressive increase in the cumulative incidence of all-cause mortality with quartiles of the air pollution PCA score (p < 0.001, Supplementary Figure S3). A significant linear association was thus found between air pollution exposure level and all-cause mortality, both in the analyses using the air pollution PCA score and considering each of the three main pollutants individually. An IQR increment in the air pollution PCA score was associated with an approximately 8.2% increase in the hazard of all-cause mortality (HR, 1.082 [95% confidence interval, 1.057–1.104]). This association appeared to be more pronounced with a rise in PM_2.5 levels, where a 1 μg/m³ elevation was associated with an excess hazard of all-cause mortality estimated at HR, 1.027 [95% CI, 1.020–1.034]. These results remained consistent regardless of the exposure lag, ranging from 1 to 3 years (Figure 2).

Similarly, a significant association was observed between air pollution and both the risks of cardiovascular and infectious causes of mortality. Notably, the association appeared to be stronger for infectious cause mortality compared to cardiovascular mortality, as evidenced by HRs of 1.686 [95% CI, 1.470–1.932] and 1.212 [95% CI, 1.074–1.368], respectively, for an IQR increment of the air pollution PCA score (Supplementary Table S4). Consistent findings were observed in the analyses focusing on individual pollutants, wherein the risk of infectious mortality exhibited a higher magnitude than that of cardiovascular mortality. Specifically, HR demonstrated an elevated risk of both cardiovascular and infection-related mortality for PM_2.5 and PM₁₀ exposure, while statistically significant associations were found between NO₂ levels and infectious mortality only.

The analysis of interactions yielded consistent results across different subgroups of the population: all-cause and cause-specific mortality appeared significantly associated with the air pollution score (Figure 3). However, these associations appeared slightly stronger in women and non-diabetic patients with minimal or no cardiovascular comorbidities. Conversely, the observed associations were not influenced by the dialysis care model, smoking status, or history of chronic respiratory disease.

The results remained largely consistent in sensitivity analyses involving alternative composite air pollution scores (Supplementary Figures S4–S6). The air pollution rank score highlighted a linear association between the risk of all-cause mortality and infection-related deaths.

This study is the first to our knowledge to demonstrate an increased risk of all-cause mortality for the two main causes of death in dialysis patients (cardiovascular and infection-related), in association with multiple air pollutant exposures. We thus observe an 8.2% increased hazard of all-cause mortality for each IQR increment of the combined air pollution PCA score. This association seems even stronger for deaths from infectious causes, with an estimated 68.6% excess hazard for each IQR increase in air pollution exposure. These results are globally consistent, regardless of the exposure lag and the alternative composite air pollution scores. Finally, certain patient profiles appear to be more vulnerable to the effects of environmental exposure, notably women and less comorbid patients.

Based on the recent literature, we were able to identify overall 10 studies investigating the association between air pollution exposure and dialysis mortality over the last decade, with variable designs, population sizes, and characteristics (10, 30–38) (Supplementary Table S5). None of these studies were conducted in Europe, the vast majority having been carried out in the USA and Taiwan. Of these studies, only six investigated the link between long-term exposure to air pollution and the risk of death in dialysis, with none considering multi-exposure (analyses mainly based on PM_2.5 exposure) (10, 30–32, 36, 38). Despite the potential for publication bias, all these studies seem to agree on a significant association between air pollution and dialysis mortality. In the study carried out by Xi Y. et al. in 2022 on a population with characteristics similar to our own (314,079 US patients, mean age 64 years), the authors highlighted a comparable effect size, with an excess hazard of death from cardiovascular causes of approximately 2% per 1 μg/m³ increment of PM_2.5 estimated at the dialysis center’s ZIP code (10). Our nationwide study thus stands out for its estimation of long-term multi-exposure to air pollution combining the three main pollutants at the patients’ residential place. These results are reinforced by the different approaches used to build the composite environmental scores, providing overall consistent results.

Another unique aspect of our research revolves around the analysis of both specific cardiovascular and infectious-related mortality, which accounts for more than one-third of deaths within our study population. The link between air pollution and cardiovascular events, all-cause mortality, and cardiovascular mortality now appears to be well-established in the literature, including in the general population (39). The dialysis patient population, despite an already increased risk of cardiovascular events due to kidney failure, also appears to be concerned by the potential influence of environmental exposure on the risk of cardiovascular events (10, 33). In their study, Xi Y. et al. also concluded that each 1 μg/m³ greater annual average PM_2.5 was associated with a 2% greater rate of cardiovascular events in US adult kidney failure patients initiating hemodialysis (10). Numerous pathophysiological explanations have already been suggested, including pro-inflammatory state, endothelial dysfunction, increased blood viscosity, cardiac autonomic system dysregulation, and other potential mechanisms inducing direct ischemia or arrhythmogenesis (39–41). One of the original features of our study lies in the identification of an additional risk of death from infectious causes, which, to our knowledge, has not been previously reported. In our study, we observed a marked excess risk of death from infectious causes, with a size effect appearing much greater than that observed for cardiovascular mortality. This original result could be explained by the potential immune dysfunction related to air pollution, in an already vulnerable population subject to relative immunodepression due to kidney failure (4, 42). Indeed, patients with kidney failure display impaired naïve and acquired immune systems related to multifactorial causes such as uremic toxins, malnutrition, and chronic inflammation, contributing to a high prevalence of infections and making sepsis the second leading cause of mortality in patients undergoing dialysis (43). Recent literature appears to highlight a strong association between air pollution exposure and infectious-related mortality: a recent study based on a cohort of over 53 million US Medicare beneficiaries observed a 9% increase in sepsis-related mortality linked to a 10 μg/m³ rise in the 12-month moving average of PM_2.5 (44). As also suggested by the recent results of the ELAPSE multicenter project, which involved over 300,000 people among six European countries, long-term exposure to air pollution was associated with an increased risk of pneumonia-related mortality (45). This may explain our results since bloodstream and respiratory infections represent two major causes of infection in the dialysis patient population (46, 47).

Our findings also suggest larger magnitudes of association between sub-chronic air pollutant exposure and health outcomes among women, and patients who were not diabetic or had fewer cardiovascular diseases at dialysis initiation. A trend, albeit not statistically significant, was also noted toward a more pronounced effect in older adult subjects. Sex disparities in the effect of air pollution on cardiovascular disease risk have been previously documented in the literature, yet results remain inconclusive (48). Some studies suggest an increased vulnerability in women, while others suggest the opposite. No definitive conclusions can be drawn in the current case, especially given the relatively moderate difference in effect size observed in our study. Interestingly, the effect size appears to be more pronounced in patients with fewer comorbidities at the initiation of dialysis, a finding consistent with previous observations (10). One potential explanation could be the relatively limited influence of environmental exposure compared to other risk factors more strongly associated with mortality in the most at-risk subgroups. Finally, the lack of interaction with the dialysis care model is in line with existing literature, as the association between air pollution and mortality holds in studies involving both peritoneal dialysis and hemodialysis patients (10, 32, 38).

This study does have certain limitations. First, due to their observational design, these results are subject to potential residual confounding. These results specifically concern the French population of dialysis patients, which might limit generalizability to other regions. However, these data are important as this is, to our knowledge, the first nationwide study of this kind in Europe. Furthermore, the literature indicates similar associations in magnitude in other populations exposed to comparable pollution levels, thus supporting external validity (10). Exposure data are based on annual average concentrations, providing an approximation of the municipality-level background pollution with a potential risk of exposure misclassification. Nonetheless, these air pollutants are generally regarded and regulated on a regional basis, and exposure modeling currently stands as the best available option at the national level (15, 17). The literature review carried out as part of this study (Supplementary Table S5) also highlighted that this is a relevant measurement scale in this context, assuming that patients with advanced kidney disease rarely change their place of residence. Furthermore, these data appear suitable for estimating sub-chronic exposure, as evidenced by consistent results across various temporal lags. Dialysis patients are more likely to spend time indoors, increasing their exposure to indoor pollutants. However, many studies have confirmed that indoor air quality is highly influenced by outdoor air quality (49). Therefore, it is likely that we are non-differentially underestimating the actual exposure levels and their health effects for our entire study population. Furthermore, the present data do not allow for more details on the type of cardiovascular or infectious events leading to patient deaths, warranting further investigations. Moreover, potential ethnic disparities in the air pollution effect could not be explored, as the collection of such individual data is prohibited in France. This study did not explore the assessment of risks associated with short-term combined environmental exposure and should also be the subject of future research endeavors, as well as the development of tools for assessing individual and indoor air pollution exposure.

In summary, our nationwide registry-based study provides compelling evidence of a significant association between multiple air pollutants and both cardiovascular and infectious cause mortality in incident dialysis patients. Moreover, the differential impact on infectious cause mortality underscores the need for targeted interventions to mitigate the adverse health effects of air pollution in this highly vulnerable population.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by French National Commission for Information Technology and Privacy (CNIL, N° 903188). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.

AH: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. MG: Conceptualization, Formal analysis, Methodology, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. FG: Investigation, Supervision, Writing – review & editing. BS: Supervision, Validation, Writing – review & editing. KL: Methodology, Supervision, Validation, Writing – review & editing. AC: Data curation, Formal analysis, Methodology, Software, Validation, Writing – original draft, Writing – review & editing. LD: Methodology, Supervision, Validation, Writing – review & editing. VG: Methodology, Supervision, Validation, Writing – review & editing. FB: Data curation, Methodology, Validation, Writing – review & editing. ML: Data curation, Methodology, Validation, Writing – review & editing. CC: Data curation, Methodology, Software, Supervision, Validation, Writing – review & editing. PA: Methodology, Supervision, Validation, Writing – review & editing. FO: Data curation, Formal analysis, Methodology, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.