Respiratory tract infections are a leading cause of hospitalization in solid organ transplant recipients, including kidney transplant patients. Immunosuppressive therapy and compromised pulmonary defenses significantly increase susceptibility to environmental exposures in this population
[1]. Among these exposures, ambient air pollution represents a growing concern, with mounting evidence linking it to adverse respiratory and systemic outcomes. Transplant recipients, already immunocompromised, may be particularly vulnerable to the health effects of polluted air. Recognizing these risks, the World Health Organization (WHO) released updated Global Air Quality Guidelines in 2021, setting strict 24-hour mean thresholds: PM
2.5≤15 µg/m
3, PM
10≤45 µg/m
3, O
3 (8-hour) ≤100 µg/m
3, SO
2≤40 µg/m
3, NO
2≤25 µg/m
3, and CO≤4 mg/m
3[2]. In contrast, the Chinese GB3095‑2012
[3] standard allows higher daily exposure limits: PM
2.5≤35 µg/m
3 (Grade I) and 75 µg/m
3 (Grade II); PM
10≤50 and 150 µg/m
3; SO
2≤50 and 150 µg/m
3; NO
2≤80 µg/m
3; O
3≤100 and 160 µg/m
3; CO≤4 mg/m
3.
PM
2.5 is particularly hazardous due to its ability to penetrate deep into the alveoli and enter the systemic circulation, triggering a cascade of biological effects
[4-5]. Global burden estimates attribute millions of premature deaths each year to PM
2.5 exposure, primarily through respiratory and cardiovascular pathways
[6-9]. Epidemiological studies have consistently shown that even short-term increases in PM
2.5 concentrations are associated with exacerbations of asthma
[10] and chronic obstructive pulmonary disease (COPD)
[11], increased risk of pneumonia, and other acute respiratory infections
[12]. These effects are not limited to high pollution levels—adverse outcomes have been reported even at concentrations below regulatory thresholds
[13]. Beyond respiratory morbidity, PM
2.5 exposure is also linked to cardiovascular events, including ischemic heart disease, arrhythmias, hypertension, stroke, and heart failure
[6]. Mechanistic studies
[14-16] implicate oxidative stress, systemic inflammation, endothelial dysfunction, and autonomic imbalance as key biological pathways. Transplant recipients, especially those with kidney grafts, are uniquely vulnerable to these environmental hazards
[17]. The combined immunosuppression, pre-existing comorbidities, and post-operative fragility compound their risk for pollution-related complications
[18]. Despite this, few studies have explored the short-term impacts of air pollution—particularly PM
2.5—on hospitalization outcomes such as medical costs and length of stay in this immunocompromised population.
Distributed lag non-linear models (DLNMs) offer a robust statistical framework for examining delayed and non-linear effects of environmental exposures on health outcomes
[19]. DLNMs have been widely applied to assess air pollution-related mortality and emergency visits in the general population
[20-22], but their use in evaluating post-transplant hospitalization metrics remains limited.
This study aims to investigate the short-term associations between ambient air pollutants (PM2.5, PM10, O3, SO2, NO2, CO) and hospitalization outcomes—including total cost and length of stay—among kidney transplant recipients admitted for respiratory tract infections. A DLNM approach was used to assess both lag-specific and cumulative effects of pollutant exposure during the 7 days prior to symptom onset.
1 Participants and methods
1.1 Ethics statement
This study was approved by the Ethics Committee of the Third Xiangya Hospital of Central South University (Approval number: 22294). The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. All participants were informed about the purpose of the study, assured of confidentiality, and provided written consent prior to participation. Participation was voluntary, and respondents could withdraw at any time without penalty.
1.2 Participants and data collection
We conducted a retrospective time-series study among kidney transplant recipients admitted to the Third Xiangya Hospital of Central South University between 2017 and 2023 due to respiratory tract infections. For each patient, the date of symptom onset was determined based on clinical complaints and was linked—along with the residential address—to local environmental monitoring data. In addition to environmental exposure variables, we collected detailed patient-level information, including age, sex, body mass index (BMI), education level, chief syndrome, documented complications, admission and discharge dates, and total hospitalization cost.
Patient residential addresses were first geocoded using the Google Maps API to obtain precise latitude and longitude coordinates. Daily concentrations of 6 ambient air pollutants—PM
2.5, PM
10, O
3, SO
2, NO
2, and CO—were obtained from the ChinaHighPM2.5: High-resolution and High-quality Ground-level PM
2.5 Dataset for China (2000—2023)
[23]. For each pollutant, daily exposure values from the day of symptom onset (lag 0) to 7 days prior (lag days 0 to 7) were compiled. Meteorological data, including daily mean temperature and relative humidity, were retrieved from the NASA POWER API (
https://power. larc.nasa.gov) and matched to each patient’s location and time frame.
1.3 Outcome measures
The primary outcomes were total hospitalization cost [measured in Chinese Yuan (CNY)] and length of hospital stay (in days). Hospitalization cost was treated as a continuous variable and modeled using a Gaussian family, while length of stay, characterized as count data with potential overdispersion, was modeled using a quasi-Poisson family.
1.4 Statistical analysis
Data processing and analysis were performed using Python 3.9 (with libraries such as pandas, xarray, netCDF4, datetime, openpyxl, etc.) and R 4.4.2 (including the dlnm package and other relevant packages). Continuous variables with a normal distribution are presented as mean±standard deviation. Categorical variables were presented as frequencies (percentages). DLNMs were used to characterize the association between each pollutant and hospitalization outcomes, accounting for non-linear exposure-response and lag-response relationships. The exposure-response function was modeled using B-splines, and the lag-response function was modeled using a natural cubic spline over lag days 0 to 7. For each pollutant, a cross-basis matrix was created using the crossbasis function from the R dlnm package
[19]. The regression models adjusted for potential confounders, including age, sex, BMI, natural spline functions for temperature and humidity (each with 3 degrees of freedom), day of the week (DOW), and holiday indicator.
The model for hospitalization cost was specified as a generalized linear model with a Gaussian distribution: E(Yi)=β0+cb(Pollutant 0-7,i)+β1·ageᵢ+β2·Sex i+β3·BMI i+f1(Temperature i)+f2(Humidity i)+DOW i+Holiday i. Where cb(Pollutant 0-7,i) denotes the cross-basis function capturing the non-linear and lagged effects of air pollutant exposure, and f₁ and f₂ represent natural spline functions for temperature and humidity, respectively. A similar model was fitted for hospital stay duration using a quasi-Poisson distribution to account for overdispersion. Cumulative and lag-specific RR and 95% CI were estimated for increasing concentrations of each pollutant, using the following reference values based on the Chinese National Ambient Air Quality Standards (GB 3095-2012): 35 µg/m3 for PM2.5, 50 µg/m3 for PM10, 50 µg/m3 for SO2, 80 µg/m3 for NO2, 100 µg/m3 for O3, and 1 mg/m3 for CO.
2 Results
A total of 874 kidney transplant recipients hospitalized for respiratory tract infections were included in the analysis. The age was (43.93±11.07) years, and 341 patients (39.02%) were female. The BMI was (21.84±3.24) kg/m². Among the cohort, the most common presenting symptom was cough (67.51%), followed by fever (52.75%) and chest distress (15.10%). Regarding education, the majority of patients had completed high school (55.83%), followed by university education (38.22%) and elementary school (5.95%). In terms of comorbidities, 9.73% had diabetes, 30.21% had hypertension, 2.86% had coronary heart disease (CHD), and 0.34% had a history of stroke. Most admissions occurred during winter (37.30%), followed by spring (23.11%), summer (20.02%), and autumn (19.57%;
Table 1).
On case days, the concentration of PM
2.5 was (41.29±26.02) µg/m
3, with a range of 4.40 to 185.80 µg/m
3. PM
10 was (63.69±35.62) µg/m
3, ranging from 9.70 to 241.10 µg/m
3. Among the gaseous pollutants, the values were 2.60 to 52.20 (10.31±4.88) µg/m
3 for SO
2, 0.26 to 1.77 (0.91±0.24) mg/m
3 for CO, 5.10 to 89.90 (26.51±14.21) µg/m
3 for NO
2, and 7.40 to 231.30 (88.75±40.3) µg/m
3 for O
3. Meteorological conditions also varied substantially, with a mean temperature of -4.53 to 32.94 (15.55±9.25) ℃ and a relative humidity of 42.12% to 95.56% (80.06%±10.08%;
Table 2).
Based on a reference level of PM
2.5=35 µg/m
3, we examined changes in hospitalization costs and length of stay among kidney transplant recipients who developed respiratory tract infections, across varying PM
2.5 concentrations and lag days. At lag day 0, higher PM
2.5 concentrations—particularly at 200 µg/m
3—were associated with increased hospitalization cost and duration (Figure
1A and
1B). However, isolated lag effects beyond day 0 were less pronounced (Figure
1A and
1B). In contrast, cumulative effects were more substantial. At PM
2.5=160 µg/m
3, the cumulative effect at lag day 1 showed an increase in hospitalization cost of 9 544.5 yuan (95%
CI 2 548.8 to 21 637.7), with this upward trend becoming more evident at higher concentrations (
Figure 1C). For hospitalization duration, cumulative
RRs at lag days 0 and 1 rose consistently with increasing PM
2.5 levels, whereas lag days 6 and 7 showed minimal or even inverse effects (
Figure 1D). Overall, the impact of PM
2.5 on both hospitalization cost and duration was more pronounced at higher concentrations, while short-term fluctuations within the national Grade I standard threshold (≤35 µg/m
3) had limited influence. At extremely elevated levels, such as 200 µg/m
3, exposure on lag day 0 markedly increased both clinical and economic burdens (Figure
1E and
1F).
The effects of PM
10 on hospitalization costs and length of stay were both showing a slight decrease at lag day 7 at 50-250 µg/m
3 (Figure
2A and
2B). However, when examining cumulative effects, a clearer pattern emerged: at lag day 5, elevated PM
10 levels were associated with the cost (Figure
2C and
2D) and notably increased in hospitalization duration. Specifically, at PM
10=250 µg/m
3, the
RR for hospital stay was 1.12 (95%
CI 0.34 to 3.72), and this rose to 1.63 (95%
CI 0.18 to 15.94) at PM
10=300 µg/m
3 (
Figure 2D). When considering both lag structure and exposure concentration, PM
10 values below 150 µg/m
3 appeared to have limited impact on hospitalization outcomes. However, once PM
10 exceeded the national Grade II standard (150 µg/m
3), both hospitalization cost and duration began to increase substantially (Figure
2E and
2F), except for lag 0.
The Grade I standard for SO
2 is 50 µg/m
3. When SO
2 concentrations remained below this threshold, the trends in hospitalization cost and duration showed some fluctuations (Figure
3A and
3B). However, when SO
2 reached higher concentrations—particularly above 75 µg/m
3—its cumulative effect on hospitalization duration became more pronounced (
Figure 3C), and hospitalization costs also increased across most lag days, except for lag 0 and lag 7 (
Figure 3D). When SO
2 levels were below 50 µg/m
3, the effects on both outcomes were minimal. In contrast, once concentrations exceeded this threshold, the impact on hospitalization burden became substantially more evident (Figure
3E and
3F).
For CO, NO
2, and O
3, no consistent or robust short-term associations with hospitalization cost or duration were observed. While the cumulative exposure-response curve for CO suggested a possible increase in hospitalization cost at concentrations exceeding 2.0 mg/m
3—particularly at lag days 0 and 1—the lag-specific effects remained generally flat, and the impact on hospitalization duration was limited to a delayed increase around lag day 6 (Supplementary
Figure 1,
https://doi.org/10.57760/sciencedb.xbyxb.00175). For NO
2, both lag-specific and cumulative effects showed minimal and inconsistent variation across concentration ranges, with wide confidence intervals and no clear dose-response pattern (Supplementary
Figure 2,
https://doi. org/10.57760/sciencedb.xbyxb.00175). Similarly, exposure to O
3 across a wide range of concentrations (50-250 µg/m
3) was not associated with meaningful changes in either hospitalization cost or duration (Supplementary
Figure 3,
https://doi. org/10.57760/sciencedb.xbyxb.00175). Overall, these findings suggest that short-term exposure to CO, NO
2, and O
3 may not exert a stable or clinically significant influence on hospitalization burden in this study population.
3 Discussion
This study assessed the short-term impacts of air pollution on hospitalization burden—including both cost and length of stay—among kidney transplant recipients with respiratory tract infections. Our findings highlight that PM2.5 and inhalable particles PM10, as well as SO2, have measurable and often substantial associations with increased clinical and economic burden, particularly at higher concentrations and during cumulative exposures. In contrast, CO, NO2, and O3 showed minimal or inconsistent effects.
Among all pollutants, PM
2.5 had the most consistent and significant association with hospitalization outcomes. Exposure to elevated PM
2.5 levels—especially at or above 160 µg/m
3—was linked to marked increases in both cost and length of stay. The effects were most acute on the same day of exposure (lag day 0), suggesting an immediate clinical impact. While isolated lag effects beyond day 0 were modest, cumulative exposures across lag days amplified the burden. These findings align with prior research showing that PM
2.5 penetrates deeply into the alveoli and exacerbates inflammation, posing a severe risk for immunocompromised individuals
[24]. Furthermore, PM
2.5 exposure is not uniform across vertical spaces. A recent study
[25] has shown that outdoor PM
2.5 concentrations peak around the 4th floor, then gradually decrease with increasing height, suggesting a vertical stratification likely influenced by traffic emissions and atmospheric mixing dynamics. For transplant recipients who spend extended time indoors, floor height may influence exposure risk, making it a meaningful factor in housing decisions.
Similarly, PM
10 demonstrated a dose-dependent cumulative effect, particularly on hospital stay duration. Short-term associations at individual lags were less consistent, but prolonged hospitalization became evident at PM
10 levels exceeding 150 µg/m
3. This suggests a threshold phenomenon where risk escalates beyond regulatory limits. Given PM
10’s tendency to irritate airways and impair mucociliary function
[26], its impact may be magnified in post-transplant individuals with reduced respiratory defenses. Individual-level behaviors also interact with environmental conditions. Moderate physical activity has proven immune benefits for transplant recipients
[27], but the opportunity to exercise safely depends heavily on environmental infrastructure. Access to urban green spaces and walkable neighborhoods not only supports physical health but may also mitigate air pollution effects.
For SO
2, our data showed relatively minor effects at concentrations under 50 µg/m
3. However, when exposure exceeded 75 µg/m
3, both hospitalization costs and duration rose significantly, especially under cumulative exposure. This aligns with SO
2’s biological potential to trigger inflammation and bronchospasm over time
[28]. Notably, in densely populated or industrial urban zones, SO
2 spikes may coincide with winter heating and poor ventilation—conditions under which respiratory infections are already more frequent, as reflected in our finding that most hospitalizations occurred in colder months.
In contrast, CO, NO2, and O3 did not show consistent or robust associations with hospitalization burden in this cohort. Though CO exposure above 2.0 mg/m3 showed a potential cost increase on early lag days, wide confidence intervals and inconsistent lag patterns suggest a limited role. O3 and NO2 exposures also varied by height and wind conditions, but their influence on outcomes here was minimal, possibly due to limited variation in individual exposure or differences in biological pathways.
From a practical standpoint, transplant recipients should adopt proactive strategies—particularly during winter and high-pollution periods. Avoiding crowded indoor spaces, firework-heavy holidays such as Lunar New Year, and exercising during low-pollution hours (such as late morning) are all advisable. Living on mid-level floors (such as 6th-11th floors) may reduce exposure to both traffic-related PM and high-altitude ozone or inversion-trapped pollutants. Additionally, the use of air purifiers and fresh ventilation systems can dramatically lower indoor PM2.5 levels, improving both comfort and clinical outcomes. Most importantly, patients and caregivers should be trained to recognize early symptoms of infection—fever, cough, and chest tightness—so that timely treatment can prevent clinical deterioration and reduce the burden of hospitalization.
Nevertheless, this study has several limitations. We were unable to account for indoor air quality, medication adherence, and social support, all of which can influence clinical outcomes. Future research should incorporate multi-pollutant models, account for regional infrastructure, and examine how chronic pollutant exposure affects long-term graft survival and quality of life.
Taken together, these results suggest that particulate pollution especially PM2.5, PM10 and SO2 poses the greatest short-term threat to post-transplant patients hospitalized with respiratory infections, with implications for both public health policy and clinical risk mitigation. Strategies targeting reductions in PM2.5 and PM10 exposure—particularly during high-pollution periods—may help reduce healthcare utilization and costs in this high-risk group. Future studies should explore mechanistic links and assess whether pre-emptive interventions such as air filtration and early outpatient monitoring during peak pollution days can mitigate adverse outcomes.
©Journal of Central South University (Medical Science). All rights reserved.
,
京公网安备11010202008535号
PDF
(2233KB)
