Associations of long-term exposure to ambient sulfur dioxide, carbon monoxide, ozone, and benzene with risk of incident chronic kidney disease in the UK

Limited studies have examined associations of gaseous air pollutants exposure with chronic kidney disease (CKD) in Europe. This study aimed to calculate the relationships between long-term exposure to ambient sulfur dioxide (SO2), carbon monoxide (CO), ozone (O3), and benzene and CKD in the UK. We included 502 369 participants from the UK biobank cohort. Associations of SO2, CO, O3, and benzene with CKD were estimated using Cox proportional hazards model. The shape of the exposure-response association between each air pollutant and CKD was then depicted using the shape constrained health impact function. We finally estimated the incidence of CKD attributable to each air pollutant by linking the constructed exposure-response association to the 2019 Global Burden of Disease data. Our results suggested SO2, high O3 days (daily max 8 hr O3 concentration > 120 µg m−3), CO, and benzene were positively associated with the risk of incident CKD. The hazard ratios (HRs) of CKD for SO2, CO, and benzene were 1.058 (95% CI: 1.039–1.078), 1.003 (95% CI: 1.001–1.005), and 1.619 (1.433–1.829) for every 1 μg m−3 increase in the concentration, respectively. For high O3 days, the HR of CKD was 1.044 (95% CI: 1.032–1.056) for every 1 d increase, but correlation to O3 concentration did not reach the statistical significance in the time-varying model. The risk of CKD increased non-linearly with increasing SO2, high O3 days, and CO, and linearly with increasing benzene. We estimated that 7.9%, 16.0%, 8.0% of incident CKD cases in the UK in 2021 could be attributed to exposure to SO2, O3, and benzene, respectively. We concluded that exposure to SO2, CO, O3, and benzene were all positively associated with increased CKD risk. Our findings highlight the importance of considering air pollution while making strategies targeting on CKD management.


Introduction
Chronic kidney disease (CKD) has gradually been recognized as one of the leading public health problems worldwide.Its global prevalence is estimated to be 9.1%-13.4%,equivalent to approximately 700 million to one billion individuals worldwide [1,2].CKD is a disease characterized by irreversible and progressive deterioration of renal function.Despite recent advances in dialysis therapy, comorbidity and mortality of CKD patients remain high, especially in those with end-stage renal disease (ESRD) [3].Understanding the risk factors that may influence incident CKD is key to mitigating the increasing disease burden.
Since 2017, a growing body of literature suggested long-term exposure to particulate matter (including PM 2.5 [4,5] and PM 10 [5,6]), as well as gaseous air pollutants including carbon monoxide (CO) [6], nitrogen oxide (NO X ) [6,7], sulfur dioxide (SO 2 ) [8], and ozone (O 3 ) pollution [9] were associated with an increased risk of CKD and/or progression.Possible mechanisms of kidney damage associated with air pollution include oxidative stress and inflammation, vascular endothelial injuries, formation of immune complexes, as well as circulating autoantibody against phospholipase A2 receptor [10].However, there were also quite a few studies that did not support these positive associations.For example, a cohort study of 24 407 individuals in South Korea found that longterm exposure to PM 10 , SO 2 , and CO were significantly associated with increased risk of CKD in the unadjusted model only-all significant associations disappeared after adjusting for covariates [11].Similarly, one cross-sectional study in mainland of China suggested that long-term exposure to SO 2 and O 3 was not significantly associated with CKD prevalence [12].
In addition to conflicting findings in existing literature, earlier studies have been geographically concentrated in East Asia and North America, which limits the generalizability of the findings to population with diverse socio-demographic and ethnical backgrounds.Two recent studies investigated the association of CKD with long-term exposure to PM and NO X in Sweden and the UK [7,13], but less is known about the role of other major gaseous pollutants such as CO, SO 2 , and O 3 in Europe.Moreover, most exposure-response relationships have been from areas with moderate concentrations of air pollutants (e.g.South Korea and Taiwan of China) and provided little evidence of public health implications for areas with very low-exposure levels.
Hence, given the conflicting conclusions of current studies and gaps in existing knowledge, we have conducted a comprehensive cohort study using the UK biobank data to examine the long-term associations between CKD incidence and specific air pollutants (CO, SO 2 , O 3 , and benzene).We also depicted the exposure-response curve for each air pollutant and further estimated the CKD burden attributed to these air pollutants in the UK.

Study population
This study included 502 369 participants recruited by the UK biobank from England, Wales, and Scotland.
Baseline questionnaire and anthropometric measurements were completed in 2006-2010 to collect information on socio-demographic characteristics, lifestyle, and environmental factors.Ongoing followup was conducted through linkage to their primary care, hospital inpatient, cancer, and death records [14].The complete study protocol is described elsewhere [15].UK biobank obtained ethical approval from the North West-Haydock Research Ethics Committee (REC reference: 21/NW/0157) and conducted research in accordance with the Declaration of Helsinki.This study was approved by the UK biobank under application 90 018.

Ambient air pollution
Annual average concentration data of SO  [16].High O 3 days is defined as the number of days on which the daily max 8 hr concentration is greater than 120 µg m −3 .The 120 µg m −3 derived from one of the target values for O 3 concentrations in the Air Quality Standards Regulations 2010-A maximum daily 8 hr mean concentration of 120 µg m −3 , not to be exceeded on more than 25 d per calendar year averaged over three years [17].Additional data of O 3 concentration in 2003-2021, with a coarse resolution of 0.75 • × 0.75 • however, were obtained from Copernicus Atmosphere Monitoring Service (CAMS) global reanalysis (EAC4) monthly averaged fields [18].

CKD assessment
Incident CKD stages 3-5 or renal failure after the baseline (i.e.recruitment date) were determined by inpatient International Classification of Disease version 10 (ICD)-10 codes (N18.0,N18.3, N18.4,N18.5, N18.8, N18.9, I12.0, I13.1, and I13.2, see detailed descriptions for these codes in supplementary table 1).Prevalent CKD at or before the baseline with ICD-10 diagnoses and/or with estimated glomerular filtration rate (eGFR) <60 ml min −1 /1.73 m 2 and/or urine albumin to creatinine ratio >300 mg g −1 were excluded from the analysis.An additional report for algorithmically-defined ESRD (Data-Field 42 026 in UK biobank) derived from three sources including baseline interview, hospital admission data, and death register data was also used to identify any cases missed by inpatient ICD codes.

Covariates assessment
Covariates including age, sex (female or male), annual household income (<£18 000, £18 000-30 999, £31 000-51 999, £52 000-100 000 or ⩾£100 000), current employment status (nonemployed or employed), qualifications (college/university or others), ethnicity (non-white or white), activity (moderate/vigorous activity or others), body mass index (BMI, <25, 25-30, ⩾30 kg m −2 ), smoking (never and ever), drinking (never and ever), and eGFR (⩾90 or <90 ml min −1 /1.73 m 2 ) were obtained at baseline.Moderate/vigorous activity was defined as participants engaging in moderate/vigorous activities for more than 10 consecutive minutes on more than 4 d/week.BMI was calculated as weight (in kilograms) divided by height (in meters) squared, with weight and height measured by trained nurses at the assessment center.Comorbidities including hypertension (yes or no), diabetes (yes or no), dyslipidemia (yes or no), and cardiovascular disease (CVD, yes or no) were extracted using the ICD-10 codes from the First occurrences data (under Category 1712 in UK biobank, see supplementary table 1 for detailed ICD-10 codes for these diseases).

Statistical model
To ensure sufficient time for the outcome to occur, we restricted participants with a follow-up period > 180 d.Follow-up time was defined from the date of recruitment to the date of first CKD onset, death, lost to follow-up, or the end of current followup (31 October 2022, 31 August 2022, and 31 May 2022 in England, Scotland, and Wales, respectively).We also excluded participants with missing data on home location or air pollutants.
Cox proportional hazards model was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for CKD associated with each air pollutant.Both time-fixed and time-varying models were fitted (only time-fixed model was fitted for CO given the data availability).In the time-fixed model, we used air pollution data in the baseline year of 2010 and matched them with the health records of included participants by the geographical coordinates of home location at assessment.In the time-varying model, air pollution was included as a time-varying exposure and matched with health records by both home location and the calendar years.We split the follow-up time into multiple one-year intervals, and calculated moving average concentrations in the corresponding calendar year and the preceding 1-4 years (hereafter referred to as the 2 to 5 year moving average).We repeat the analysis four times using 2 to 5 year moving averages, and the model with the highest HR was selected as the main model.All covariates we mentioned earlier were adjusted in both time-fixed and time-varying models.In time-varying model, in addition to air pollution, we also assumed age and comorbidities (hypertension, diabetes, dyslipidemia, and CVD) were time-varying variables whereas all other variables were time-fixed because these data for the entire sample were collected only once at baseline.For each covariate with missing data, a column of binary indicators was added showing which participant has missing data and which does not, and both the raw data and the indicator were included in the model.HRs and 95% CIs for CKD associated with every unit (1 µg m −3 or 1 d) increase in air pollutant measures, which represented an overall effect, were reported to allow for better comparison with previously published studies.We also reported HRs and 95% CIs for the 2nd to 4th quartiles (Q2-Q4) compared with the 1st quartile (Q1) for each air pollutant.
To ensure the robustness of the estimated HRs, we performed four sensitivity analyses of the timevarying models (supplementary methods).Subgroup analyses for the time-varying model were also conducted to test differences in CKD risk according to age (<65 and ⩾65 years), sex (female and male), smoking (never and ever), and genetic risk (high and low).We specially selected smoking status and genetic factors because of their potential to interact with air pollutants to influence CKD or other health events such as CVD [13,19].We used the enhanced polygenic risk score (PRS) for creatinine-based eGFR provided by the UK biobank to represent the genetic risk for CKD [20].Enhanced PRS was calculated for 104 231 individuals in the UKB testing subgroup, in which the PRS algorithm has been trained on both external data and an additional training data from a subset of UKB data [20].Given that eGFR values are inversely related to CKD risk, we set PRS ⩾ median and <median to low and high genetic risk of CKD, respectively.In each subgroup analysis, the same set of covariates as the full model were adjusted, and the first 10 principal components of ancestry and genotyping batch were additionally adjusted in the subgroup analysis by genetic risk.Between-group differences were examined by using the Z-test [21].
We depicted the shape of the exposure-response association between air pollutants and CKD using the shape constrained health impact function (SCHIF) [22].SCHIF uses Cox regression model to capture a variety of potential nonlinear functions and then calculates an ensemble model that takes into account all fitted shapes.SCHIF shapes are smoother and less likely to be affected by sparse data than natural cubic splines [22].
We also calculated the incidence of CKD in 2021 attributed to air pollutants that were positively association with CKD.The latest CKD incidence rate in the UK was obtained from Global Burden of Disease 2019 [23].We assumed that (1) the incidence rate in 2021 would be similar to that in 2019 as it has remained relatively stable after 2015 [23] and (2) the incidence rate at national level would be the same across all grid cells in the UK.Based on the exposureresponse association calculated in the main timevarying model, we calculated attributable fraction (AF) at each 1 km × 1 km grid cell of the UK using an adapted equation AF = (HR−1)/HR [6,24].Here, we used HR as a surrogate for relative risk given the incidence of CKD in 2019 in the UK was relatively low (0.22%) [23,24].The minimum concentration of each air pollutant was used as the theoretical minimum risk exposure level.Latest concentrations of SO 2 , high O 3 days, CO, and benzene in 2021 across the UK from Defra were used.O 3 concentration of 2021 were obtained from CAMS European air quality reanalysis in EAC4.Gridded population data was obtained from the Environmental Information Data Centre [25].We obtained 95% Uncertainty Interval (UI) of the attributable burden based on the 95% CI of HR from SCHIF.R version 3.6.3(R Foundation for Statistical Computing) was used in all statistical models.Cox regression models were constructed using the 'survival' package.SCHIF was constructed using the 'bestcox' function written by Nasari et al [22].A twosided P value < 0.05 was regarded as statistically significant.

Participants characteristics in UK biobank
Participants were excluded based on pre-defined exclusion criteria: 9462 with pre-existing CKD diagnoses, 483 due to follow-up durations of ⩽180 d, 211 for missing home address, and 41 for absence of air pollutant data.Hence, the final sample included 492 203 participants for analysis of O 3 concentrations, and 492 162 individuals for analysis of remaining air pollutants including SO 2 , high O 3 days, CO, and benzene (supplementary figure 1).Over a median follow-up period of 13.6 years, 18 503 participants were diagnosed with incident CKD, accounting for 3.7% of the total cohort (table 1).The median age of the sample was 58.0 years, and the proportion of men was 45.5%.Median concentrations of SO 2 , O 3 , CO, and benzene at baseline were 2.19 µg m −3 , 39.4 µg m −3 , 220.01 µg m −3 , and 0.5 µg m −3 , respectively.Median number of days on which the daily max 8 hr O 3 concentration is greater than 120 µg m −3 was 1.14.Over the follow-up, concentrations of multiple air pollutants excluding O 3 showed a declining trend (supplementary figures 2-6 and table 2).

Association between CKD and air pollutants
In multivariate-adjusted Cox regression models, we observed that SO 2 , CO, high O 3 days, and benzene exposure were all positively associated with CKD in the time-fixed model and/or the time-varying model (table 2).Missing indicators for all covariates yielded no statistically significant results, suggesting an unbiased nature of the missing data and thereby confirming the reliability of the risk assessments.In timevarying models, optimal models were selected based on the highest HRs, which suggested using 2 year moving average for SO 2 and O 3 concentration and 5 year moving average for high O 3 days and benzene (supplementary figure 7).In the time-varying models, the HRs of CKD for SO 2 and benzene were 1.058 (95% CI: 1.039-1.078)and 1.619 (1.433-1.829)for every 1 µg m −3 increase in concentration.Similarly, for every 1 µg m −3 increase in concentration of CO, the HR is 1.003 (95% CI: 1.001-1.005) in the time-fixed model.O 3 exposure showed mixed results (table 2).High O 3 days were significantly associated with increased CKD risk in both time-fixed (HR: 1.036, 95% CI: 1.011-1.061)and time-varying (HR: 1.044, 95% CI: 1.032-1.056)models, but correlation to O 3 concentration did not reach the statistical significance in the time-varying model.Sensitivity analyses were broadly consistent with the main results, albeit with slight variations in HR estimates (supplementary table 3).
We conducted subgroup analyses according to age, sex, and smoking status in 492 162 participants (492 203 for O 3 concentration), and according to genetic risk in 102 260 participants.We found that participants aged ⩾65 years were more susceptible to the negative impact of high O 3 days on CKD risk compared with those aged <65 years (p for difference: 0.02, figure 1).We also observed slight differences in the risk of O 3 concentration on CKD between female and male populations (p for difference < 0.01), but both groups have shown a borderline CI (p for female = 0.047, p for male = 0.048).We did not observe statistically significant differences between other subgroups for the four air pollutants.

Exposure-response association between CKD and air pollutants
Using SCHIF, we estimated the exposure-response relationship between each air pollutant and CKD within the UK.HR for CKD showed a sublinear association with CKD over lower concentration and then a near-linear association over high concentrations of SO 2 .This finding remained consistent across the time-fixed and time-varying models (figure 2 and supplementary figure 8).For high O 3 days, HR also showed a sublinear association with CKD over lower number of days and then a near-linear association over higher number of days in time-varying model, whereas a different pattern was shown in time-fixed model that increased risk of CKD was only observed at high number of days.HR of CKD showed a linear association over the entire benzene range in both time-fixed and time-varying models.For CO, HR of CKD increased supra-linearly over lower concentration and then increased linearly in time-fixed model.We did not observe statistically significant exposureresponse associations of O 3 concentration with CKD in either time-varying or time-fixed models.

Attributable CKD incidence to air pollutant
We further predicted incident CKD cases in the UK attributed to air pollutants.A total of 10 770.91 Covariates including age, sex, ethnicity, household income, qualifications, current employment status, smoking, drinking, activity intensity, BMI, hypertension, diabetes, dyslipidemia, and CVD were adjusted in both time-fixed and time-varying model.In time-fixed model, we used air pollution data in 2010 as the exposure.In time-varying model, we used 2 year moving average for SO2 and O3 concentration and 5 year moving average for high O3 days and benzene as the exposure, respectively.

Discussion
In this study, we examined the impact of four air pollutants (SO 2 , O 3 , CO, and benzene) on the risk of incident CKD in the UK biobank cohort.Incident CKD cases attributed to air pollutants were also estimated in the UK based on constructed exposure-response associations.Our results suggested SO 2 , high O 3 days (daily max 8 hr O 3 concentration > 120 µg m −3 ), CO, and benzene were positively associated with CKD.We estimated that 7.9%, 16.0%, 8.0% of incident CKD cases in the UK in 2021 could be attributed to exposure to SO 2 , O 3 and benzene, respectively.There have been some previous literatures on the relationship between long-term exposure to SO 2 and CKD-mostly from China and South Korea, and the findings were inconsistent [26].Our study represented the first study outside of Asia to show a positive association between long-term SO 2 exposure and CKD, which was in line with several previous studies [8,[27][28][29].For example, a large cohort study in Taiwan of China suggested that long-term exposure to SO 2 was associated with a 1.026-fold (95% CI: 1.019-1.029)increased risk of CKD for every 1 µg m −3 [1/2.661= 0.376 parts per billion] increase [8], which was comparable to our estimated HR (1.058, 95% CI: 1.039-1.078).Some other studies-most were cross-sectional studies, however, refuted this positive association [11,12,[30][31][32].Hence, to further verify the association between SO 2 and the incidence of CKD, more high-quality large-scale studies, especially cohort, from multi-locations in the world are needed in future research.
We linked long-term CO exposure to a high risk of incident CKD, which also served as a complement to conclusions of previous studies [6,[27][28][29].One large cohort study in the US suggested that for every 1 µg m −3 [1/(1.16* 10 3 ) = 0.863 * 10 −3 parts per million] increase in CO concentration in the US, the HR  for CKD was 1.00034 (1.000 28-1.00038) [6], which was smaller than that (HR, 1.003, 95% CI: 1.001-1.005) in our study.This may be due to slightly higher CO concentration in the US [Median (IQR, µg m −3 ): 591.6 (464.0-742.4) vs. 220.06(214.7-226.3) in the UK], given the exposure-response plot we constructed suggested the association of CO with the risk of CKD was more significant at lower concentration (i.e.HR first increased supra-linearly at lower concentrations and then increased linearly with a small slope).
Few studies have focused on the association between benzene exposure and the incidence of CKD, and our study was the first to show that for every 1 µg m −3 increase in benzene concentration, the risk for CKD increased approximately 1.6-fold.Additionally, the exposure-response function suggested HR of CKD showed a linear association across full range of benzene in both time-fixed and timevarying models.Previous studies have shown that benzene exposure increases the risk of lung cancer [33], myocardial infarction [34], hypertension [35], and acute myeloid leukemia [36].Our results further demonstrated the multi-organ toxicity of benzene as an air pollutant.
Interestingly, for the association between the incidence of CKD and O 3 , we obtained conflicting results for two different O 3 exposure indices.Our models revealed statistically significant positive associations between high O 3 days and the incidence of CKD, but not for O 3 concentration.It is worth noting that the median concentration of O 3 in the UK was 39.4 µg m −3 , while high O 3 days was defined as days with a daily max 8 hr O 3 concentration >120 µg m −3 .Therefore, contradictory findings may suggest that a higher O 3 concentration has a greater impact on the risk of CKD.Moreover, the time-fixed model has revealed a protective effect of O 3 concentration on CKD.We assumed this may be because the coarseresolution O 3 concentration data in one single year cannot well distinguish the actual exposure levels of the included participants.Previous studies have also yielded inconsistent results regarding the association between long-term O 3 exposure and CKD.A nationwide cross-sectional study in China demonstrated that long-term exposure to O 3 may increase risk of CKD [9], whereas studies from South Korea, the US (specifically in African Americans), and other parts of China indicated no increased risk of CKD after long-term exposure [12,28,30,31,37].Two recent cohort studies have even identified a negative association between O 3 and CKD [29,38].These results imply that the impact of O 3 on the risk of CKD incidence may vary among different ethnic or regional populations.
Our subgroup analyses suggested that individuals aged ⩾65 years were more susceptible to the adverse effects of O 3 exposure on CKD.Older individuals often show signs of renal aging manifested by structural changes, declining function, and increased oxidative stress and inflammation [39], thus potentially increasing their vulnerability to air pollutants.However, our study contradicted one previous study which found a higher risk of CKD associated with O 3 exposure in younger population, and have attributed this to a protective effect of medication use in elders [9].In addition, our study showed a higher risk of CKD for O 3 concentration in women than in men.This also contradicted previous studies which showed the effect of O 3 on CKD were higher in men [28] or similar between men and women [9,38].Given these conflicting findings, more research is needed to identify susceptible populations at risk for O 3 -related CKD.
The biological mechanisms underlying the association between gaseous air pollutants and kidney function remain incompletely elucidated, but current experimental studies have provided some plausible biological evidence.As the primary organ that eliminates wastes from the body, the kidney is particularly susceptible to the effects of environmental toxins [40].For example, previous animal studies suggested that inhaling of SO 2 can lead to renal tissue lesions, induce oxidative stress responses, and cause dose-dependent increases in DNA injuries in kidney [40][41][42].In addition to inflammation responses, one study found that O 3 inhalation can modify the gene expression in rat's kidney, e.g. the increased endothelin-1 expression was found to cause endothelial dysfunction in kidney [43].Meanwhile, O 3 -induced increases in expression of serum-and glucocorticoid-inducible kinase mediated the patho-or physio-logical processes in the kidney and further affected its function [43].For CO, few studies investigated the mechanism of kidney injury associated with inhaled CO directly, but some studies suggested that exposure to CO can increase the risk of CVD (e.g.myocardial ischemia and ischemic heart failure) via increasing carboxyhemoglobin level or mediating oxidative stress pathway [44].Finally, for benzene, one study has reported that inhalation of benzene from the solvent mixture resulted in significant dose-dependent biochemical and functional changes in kidney [45].
Our findings had important public health implications and highlighted the importance of considering air pollution while making strategies targeting on CKD management.In our study, we predicted that 7.9%, 16.0%, 8.0% incident CKD cases in the UK in 2021 were likely associated with exposure to SO 2 , O 3 , and benzene respectively.In view of the high socioeconomic and health burden of CKD and excess risk of morbidity of CKD related to air pollution [1,46], reducing air pollution emission from specific sources including power station, industry, and/or traffic (for SO 2 , benzene, and precursors of O 3 ), or reduce exposure by spending less time outdoors in summer (for O 3 ) may help minimize CKD burden and improve the quality of life.
Our study has limitations.First, we adjusted for multiple known confounders in our model but effects due to other measured/unmeasured confounders might exist and influence the results and interpretation, including any interactions between air pollutants and confounders.Second, the UK biobank study did not evaluate air pollution exposure in indoor environment at home/work, potentially introducing bias on the estimated risk.Third, we focused on the association between air pollutants and CKD incidence in the UK, making it difficult to generalize our findings to populations in other regions of the world.Future research including a larger population size and a wider range of concentrations would help enhance the credibility of the results.
In conclusion, our study identified significant positive associations between CKD onset and certain air pollutants (SO 2 , CO, O 3 , and benzene).Importantly, we predicted quite a few incident CKD cases can be attributed to exposure to these air pollutants.Mitigation strategies taking in account reducing emission and exposure to these specific air pollutants might help reduce the risk of CKD.

Figure 2 .
Figure 2. Exposure-response associations between air pollutants and chronic kidney disease in the UK in the time-varying models.(a) SO2, (b) O3, i.e. high O3 days, which is defined as the number of days on which the daily max 8 hr concentration is greater than 120 µg m −3 , (c) O3 concentration, (d) Benzene.Dark blue lines represent the ensembled hazard ratios and blue shaded areas represent their 95% confidence intervals.The minimum concentration in the UK was set as the reference for each air pollutant.

Figure 3 .
Figure 3. Attributable incidence of chronic kidney disease to air pollutants in the UK.(a) SO2, (b) O3, i.e. high O3 days, which is defined as the number of days on which the daily max 8 hr concentration is greater than 120 µg m −3 , (c) Benzene.

Table 1 .
Baseline characteristics of included participants in UK biobank.O3 concentration data from EAC4 (with a coarse resolution 0.75 • × 0.75 • ).Hazard ratios and 95% CIs for chronic kidney disease associated with air pollutants in the time-varying models, stratified by age (a), sex (b), smoking (c), and genetic risk (d).O3, i.e. high O3 days, which is defined as the number of days on which the daily max 8 hr concentration is greater than 120 µg *(95% UI: 6658.743-14592.02),21803.79 (95% UI: 17 401.58-25962.06),10913.19 (95% UI: 7621.192-13812.13)incidentCKD cases were predicted to be associated with exposure to SO 2 , high O 3 days, and benzene, respectively, accounting for 7.9%, 16.0%, 8.0% of all incident CKD cases in the UK in 2021.Our data further highlighted geographical hotspots for CKD incidence attributable to these specific pollutants, particularly in large urban centers like London, Birmingham, Manchester, and Liverpool (figure3).Figure 1.

Table 2 .
Hazard ratios and 95% CIs for chronic kidney disease associated with air pollutants in the UK.
NA, not applicable.# Number of days on which the daily max 8 hr concentration is greater than 120 µg m −3 .* O3 concentration data from EAC4 (with a coarse resolution 0.75 • × 0.75 • ).