Impact factor (WEB OF SCIENCE - Clarivate)

2 year: 7.664 | 5 year: 7.127

Articles

The impact of ambient air pollution on hospital admissions, length of stay and hospital costs for patients with diabetes mellitus and comorbid respiratory diseases in Panzhihua, Southwest China

Xianzhi Li1,2,3*, Bin Yu4*, Yajie Li5*, Haorong Meng6, Meiying Shen7, Yan Yang1,3,8, Zonglei Zhou9, Shunjin Liu1,2,3, Yunyun Tian2,3, Xiangyi Xing1,3,10†, Li Yin1,2,3

1 Meteorological Medical Research Center, Panzhihua Central Hospital, Panzhihua, Sichuan Province, China
2 Clinical Medical Research Center, Panzhihua Central Hospital, Panzhihua, Sichuan Province, China
3 Dali University, Dali, Yunnan Province, China
4 Institute for Disaster Management and Reconstruction, Sichuan University – Hong Kong Polytechnic University, Chengdu, Sichuan Province, China
5 Tibet Center for Disease Control and Prevention, Lhasa, Tibet Autonomous Region, China
6 Yunnan Center for Disease Control and Prevention, Kunming, Yunnan Province, China
7 Nursing department, Panzhihua Central Hospital, Panzhihua, Sichuan Province, China
8 Department of Respiratory and Critical Care Medicine, Panzhihua Central Hospital, Panzhihua, Sichuan Province, China
9 Department of Epidemiology, School of Public Health, Fudan University, Shanghai, China
10 Department of Pharmacy, Panzhihua Central Hospital, Panzhihua, Sichuan Province, China
* Joint first authorship.
† Joint senior authorship.

DOI: 10.7189/jogh.13.04118

Share:

Facebook
Twitter
LinkedIn
Print Friendly, PDF & Email

Respiratory diseases (RD) and diabetes mellitus (DM) are major health concerns, which have contributed to a huge burden on society [1]. The Global Burden of Disease (GBD) Study 2019 suggested that chronic obstructive pulmonary disease (COPD), a serious chronic RD, was responsible for 2638.2 global age-standardised point prevalence and 42.5 deaths per 100 000 population [2], and the global age-standardised point prevalence and death rates for type 2 diabetes were 5282.9 and 18.5 per 100 000 [3]. Numerous studies have recently confirmed the pathophysiology of the metabolic glycaemic disorder and the RD with similar mechanism shared by the two conditions [4,5]. As a major comorbidity of several RD [6], DM affect the disease behaviour of patients with RD. Conversely, patients with DM are at high risk of asthma, COPD, etc. [7]. Moreover, the same pathophysiological mechanisms determining the major degenerative complications of DM may further lead to pulmonary function deficiency [5]. Given the heavy burden caused by RD and comorbid DM [8], it is crucial to explore risk factors that help to develop appropriate measures to reduce the prevalence of RD and DM comorbidity.

The GBD 2019 showed that the largest increases in risk exposure from 2010 to 2019 were ambient particulate matter pollution, drug use, high fasting plasma glucose, and high body-mass index [9]. Ambient air pollutants are one of modifiable risk factors for both RD and DM [10]. Epidemiological investigations have provided a large amount of evidence indicating the association between short-term and lifelong particulate matter (PM) and ozone (O3) exposure and RD [9,1113]. Previous studies also have linked air pollutants exposure with DM [14,15]. However, few studies have examined the impact of air pollution on RD and comorbid DM [16]. Furthermore, almost all of them focus on cities with large economies, and severe air pollution, such as Shanghai [17], Bangkok [18] and Beijing [19]. Studies have shown that differences in the toxic composition of air pollutants and climate conditions across regions may influence the incidence rate of RD and DM [20,21]. Thus, evidence from low-pollution areas is needed. Air pollution has recently attracted attention for its effect on hospital admissions, hospital length of stay (LOS), and hospital costs [22]. Compared to traditional indices such as mortality and morbidity, the three indices can intuitively reflect the disease burden.

Despite air quality has improved over the past few years, China still has a high RD burden caused by ambient air pollution [23], as well as increasing prevalence of DM [16]. Located in southwest China, Panzhihua is considered an ideal place to explore the health effects of air pollution in low-pollution areas due to its good air quality [24]. This study aimed to conduct a time-series analysis to estimate the association of air pollutants (including particles with an aerodynamic diameter <2.5 micron (μm) (PM2.5), particles with an aerodynamic diameter <2.5 μm (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), and ozone (O3)) with hospital admissions, LOS and hospital cost for RD-DM patients in Panzhihua. Besides, the attributable risk of RD-DM was estimated due to excess air pollution exposure using different reference concentration.

METHODS

Study area

Panzhihua, located in southwest China, covers about 7414 km2 and has 1.12 million permanent residents. With its location in the upper Yangtze River valley, the city has a typical dry-hot valleys (DHVs) climate characterised by high temperature and low humidity. In recent years, Panzhihua’s air quality has been much better than most metropolises. In spite of this, the major ambient particulate pollutants in Panzhihua like PM2.5 still exceed the concentration limits recommended by 2021 WHO’s Air Quality Guidelines (AQG) [25]. The study area included all districts and counties in Panzhihua.

Daily records of hospital admissions for RD-DM

Daily records of hospital admissions between 1 January 2016 and 31 December 2020 were collected from Panzhihua Health Information Center (http://wjw.panzhihua.gov.cn/jgsz/zsdw/). Each hospitalisation record was first extracted for the following information: ID, admission date, residential address, age, sex, LOS, hospital cost and primary diagnosis; then using hospitalisation records reshaped into time series data, we obtained RD-DM patients’ daily admissions, LOS and hospital costs. All diagnoses were coded using the International Classification of Diseases, 10th Revision (ICD-10). Patients with an ICD-10 diagnosis of both J00-J99 and E10-E14 were defined as having RD-DM.

Air pollution and meteorology data

We have acquired daily mean concentrations of PM2.5, PM10, SO2, NO2, CO, O3 from 1 January 2016 to 31 December 2020 from the Panzhihua Environmental Monitoring Center (http://sthjj.panzhihua.gov.cn/). And we obtained daily weather data from the Panzhihua Meteorological Bureau (http://sc.cma.gov.cn/ds/pzh/), including relative humidity and average temperature. We matched air pollution and meteorology data with date-specific hospital admissions.

Consumer price index (CPI)

In this study, we collected information on Panzhihua’s CPI for the period of 2016-2020 from the Panzhihua Bureau of Statistics (http://tjj.panzhihua.gov.cn/).

Statistical methods

Descriptive statistics

Mean, standard deviation (SD) and quartiles were used to describe the distribution of patient data, air pollutants, and meteorology factors. Besides, to show the daily variation in hospital admissions, LOS and hospital cost for RD-DM patients, the calendar heat map was employed. The correlation between air pollution and meteorological factors was estimated using Spearman rank correlation analysis.

Estimating associations

An over-dispersed generalised additive models (GAM) with a quasi-Poisson link was applied to estimate the associations between air pollutants and hospital admissions for RD-DM due to daily hospital admissions usually following a Poisson distribution [26]. We found daily LOS and hospital cost were approximately normally distributed based on our data; thus, a GAM with a Gaussian link was utilised to detect the short impact of air pollution on LOS and hospital cost.

Several potential confounding factors were controlled by cubic spline functions in the GAM model, including day of week (DOW), relative humidity, average temperature, public holidays and long-term trends and seasonality. According to previous research, the degree of freedom (df) was determined to be seven per year for long-term trends and seasonality [16,27] and three for mean temperature and relative humidity [28,29]. We included DOW and holidays as categorical variables in model. CPI is an index number measuring the average price of consumer goods and services purchased by households, with the percent change in the CPI commonly used as a measure of inflation [30,31]. It is suggested that the CPI is related to the hospitalisation expenditure of the population [31] and we thus considered the CPI for hospital cost.

Considering the delayed effect and acute effect of air pollution, single-day lag (from current day to seven days before: lag0-lag7) and multi-day moving average (from lag01 to lag07) was used to identify lag patterns in ambient air pollution [16]. According to previous literatures [3235], the basic single-pollutant models were used and the models were as follows:

log[E(Yt)] = βZt + ns (time, 7 x 5) + ns(Tempt, 3) + ns(RHt, 3) + DOW + Holidayt + intercept

LOSt = βZt + ns(time, 7 x 5) + ns(Tempt, 3) + ns(RHt, 3) + DOW + Holidayt + intercept

Costt = βZt + ns(time, 7 + 5) + ns(Tempt, 3) + ns(RHt, 3) + DOW + Holidayt + CPI + intercept

where E(Yt) represents the expected daily RD-MD counts on day t; LOSt is the length of hospital stay on day t; Costt is hospital cost on day t; β is the relative coefficient; Zt indicates the concentration of air pollutants on day t; ns represents natural smooth splined function; time is the days of calendar time on day t, used to control the long-term trend and seasonality of time; Tempt refers to the temperature on day t; RHt indicates the relative humidity on day t; DOW is day of the week, CPI is consumer price index.

The association between air pollution and outcome was expressed as percentage changes (%, PC) and the corresponding 95% confidence interval (CI) for hospital admissions and absolute increase and the corresponding 95% CI for LOS and hospital cost, respectively [16,36]. The calculation formula is below:

PC = [exp(β x Δc - 1] x 100%

Absolute increase = β x Δc

where β represents coefficient of air pollutants from GAM model; Δc is the unit increase number of air pollutants concentration. For PM2.5, PM10, SO2, NO2, O3, we set Δc as 10 μg/m3; for CO, we set Δc as 1 milligramme per cubic metre (mg/m3).

Estimating the attributable risk

Attributable fraction (AF) and attributable number (AN) based on the previously established GAM model were used to estimate the burden of RD-DM due to exceeding air pollutants exposure. The calculation formula is below:

where baseline patients are the admissions number on one day; i represents days when air pollution concentration is higher than the reference concentration; β is the exposure–response relation coefficient based on the GAM; ∆Ci stands for the difference between the daily concentration of air pollutants and the reference concentration. We choose different standards as reference concentrations: 1) the standard of WHO’s 2021 AQG [25]; 2) WHO reports that there is no threshold concentration for adverse health effects of six air pollutants [25,37]. Therefore, we also selected 0 μg/m3 as reference concentration.

Exposure-response relationship

We used the plot.gam() function from mgcv package to explore nonlinear relationship. Since the link function of the model must be a log function, we only examined the nonlinear relationship between air pollution and hospital admissions.

Subgroup analyses

Furthermore, stratified subgroup analysis by age (<65 vs. ≥65) and sex (male vs. female) was utilised to examine differences in effect values. And Z-test was employed to test the statistical significance of the differences between subgroups. The equation was listed below:

where Q1 and Q2 refer to the effect values of the subgroups, respectively; SE1 and SE2 refer to the standard error of the subgroups, respectively [38].

Sensitivity analysis

For the purpose of ensuring that our results are robust, we conducted the following sensitivity analyses: 1) to test the stability of the effects after other pollutants are taken into account, two-pollutant models were built; PM10 and PM2.5 do not appear simultaneously in the model considering the collinearity between the two air pollutants; 2) we changed the df of the long-term trend from 4 to 12 df per year; 3) E-value was computed to assess the unmeasured confounding bias. The results are considered robust if the risk ratio (RR) is less than the corresponding E-value [16,39,40]; 4) we used PM10 and PM2.5 from China High Air Pollutants (CHAP) data (http://www.geodata.cn), a data derived from model, to compare the results of fixed monitoring stations data with that of the model-based data.

All statistical analyses were performed using R software (version 4.2.1), and a two-sided P < 0.05 was considered statistically significant.

RESULTS

Results summary

Our results suggest that daily PM2.5, PM10, SO2, NO2, and CO contributed significantly to elevated risk of hospital admissions, LOS, and hospital cost for patients with RD-DM. Overall, 32.68% of hospital admissions for RD-DM could be attributed to NO2, with the highest population attributable faction. O3 showed protective effects on hospital admissions, LOS, and hospital cost among RD-DM patients, though two-pollutant models suggest the potential joint risk effects of O3 with other pollutants. The effects of air pollutants on hospital admissions, LOS and economic cost for RD-DM were more pronounced in people over the age of 65 and male patients than their counterparts.

Baseline characteristics

We included 17 320 RD-MD patients in our study (Figure S1 in the Online Supplementary Document), leading to 213 082 days in LOS and 207.50 million CNY in hospital costs. Daily RD-DM admissions, LOS, and hospital costs ranged from five to 29 cases, 0.00 to 676.00 days, and 0.00 to 712 021.87 CNY, respectively (Table 1). A total of 62.74% of the patients were over 65 years old. There were 10 785 (62.27%) male patients and 6535 (37.33%) female patients (Table S1 in the Online Supplementary Document). Daily mean concentrations were 28.24 μg/m3 for PM2.5, 51.88 μg/m3 for PM10, 32.05 μg/m3 for SO2, 33.54 μg/m3 for NO2, 1.39 mg/m3 for CO, 80.85 μg/m3 for O3; in terms of daily mean temperatures and relative humidity, they were 21.02°C and 58.73%, respectively (Table 1). Figure S2 in the Online Supplementary Document shows correlations between air pollution levels and meteorological variables. PM2.5 and PM10 showed the strongest correlation. The calendar heat map (Figure S3 in the Online Supplementary Document) shows the daily variation in hospital admissions, LOS and hospital cost for participants.

Table 1. Distribution characteristics of daily cases of diabetes mellitus and comorbid respiratory diseases (RD-DM), air pollutants and meteorological factors in Panzhihua from 2016 to 2020

WordPress Data Table

SD – standard deviation, Min – minimum value, P25 – first quantile, P50 – second quantile, P75 – third quantile, Max – maximum value, LOS – length of hospital stay, PM2.5 – particles with an aerodynamic diameter <2.5 micron (μm), ug/m3 – microgrammes per cubic metre, PM10 – particles with an aerodynamic diameter <10 μm, SO2 – sulfur dioxide, NO2 – nitrogen dioxide, CO – carbon monoxide, mg/m3 – milligrammes per cubic metre, O3 – ozone

Effect of air pollution on RD-DM

Figure 1 illustrates the PC value for six air pollutants with varying lag days in RD-DM hospital admissions. All air pollutants except O3 show statistically adverse effects. There was a maximum effect value at lag 07 for six pollutants, with PM2.5 being 7.25% (95% CI = 4.26 to 10.33), PM10 being 5.59% (95% CI = 3.79 to 7.42), SO2 being 10.10% (95% CI = 7.29 to 12.98), NO2 being 12.33% (95% CI = 8.82 to 15.95), CO being 25.77% (95% CI = 17.88 to 34.19), and O3 being -2.99% (95% CI = -4.08 to -1.90). Subgroup analysis shows that the effect is more pronounced in people over 65 than in young adults; similarly, the effect is stronger in male patients than in female patients (Table S2 to Table S13 in the Online Supplementary Document).

Figure 1. The effect of six air pollutants on admission with single pollutant model. Percentage change for particles with an aerodynamic diameter <2.5 micron (μm) (PM2.5), particles with an aerodynamic diameter <10 μm (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2) and ozone (O3) were per 10 µgmes per cubic metre (ug/m3) increase and 1 mgmes per cubic metre (mg/m3) for carbon monoxide (CO); younger adult: age less than or equal to 65 years old; the elderly: age over 65 years old.

Figure 2 shows the absolute increase with LOS in six air pollutants with different lag days. All air pollutants except O3 show statistically different adverse effects. For LOS, the lag day of maximum effect value slightly varies among six pollutants for the total population in which PM2.5 is 12.33 days (95% CI = 8.06 to 16.61) at lag 07, PM10 is 9.15 days (95% CI = 6.56 to 11.74) at lag 07, SO2 is 11.98 days (95% CI = 8.29 to 15.67) at lag 05, NO2 is 20.00 days (95% CI = 15.24 to 24.76) at lag 07, CO is 34.37 days (95% CI = 24.58 to 44.17) at lag 07 and O3 is -4.07 days (95% CI = -5.68 to -2.46) at lag 07. In age subgroup analysis, the effect is more pronounced in people over the age of 65 than in the young adult; similarly, the effect is stronger in male patients than in female patients (Table S2 to Table S13 in the Online Supplementary Document).

Figure 2. The effect of six air pollutants on length of hospital stay (LOS) with single pollutant model. Absolute increase for particles with an aerodynamic diameter <2.5 micron (μm) (PM2.5), particles with an aerodynamic diameter <10 μm (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2) and ozone (O3) were per 10 µgmes per cubic metre (ug/m3) increase and 1 mgmes per cubic metre (mg/m3) for carbon monoxide (CO); younger adult: age less than or equal to 65 years old; the elderly: age over 65 years old.

Figure 3 shows the absolute increase with hospital cost in six air pollutants with different lag days. All air pollutants except O3 show statistically different adverse effects. The lag day of maximum effect value slightly varies among six pollutants for total population in which PM2.5 is 12 279.18 CNY (95% CI = 7063.50 to 17 494.86) at lag 07, PM10 is 9397.81 CNY (95% CI = 6234.08 to 12 561.53) at lag 07, SO2 is 13 269.71 CNY (95% CI = 8771.65 to 17 767.76) at lag 05, NO2 is 21 551.46 CNY (95% CI = 15 750.94 to 27 351.97) at lag 07, CO is 38 540.25 CNY (95% CI = 26 593.88 to 50 486.62) at lag 07 and O3 is -4942.53 CNY (95% CI = -6906.63 to -2978.44) at lag 07. Subgroup analysis shows the effect is more pronounced in people over the age of 65 than in the young adult; similarly, the effect is stronger in male patients than in female patients (Table S2 to Table S13 in the Online Supplementary Document).

Figure 3. The effect of six air pollutants on hospital cost with single pollutant model. Absolute increase for particles with an aerodynamic diameter <2.5 micron (μm) (PM2.5), particles with an aerodynamic diameter <10 μm (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2) and ozone (O3) were per 10 µgmes per cubic metre (ug/m3) increase and 1 mgmes per cubic metre (mg/m3) for carbon monoxide (CO); younger adult: age less than or equal to 65 years old; the elderly: age over 65 years old.

Attributable risk of RD-DM due to air pollution

Table 2 summarised the attributable risk due to exceeding air pollutants exposure using the standard of WHO’s 2021 AQG and 0 μg/m3 as the reference concentrations, respectively. For hospital admissions in the total population, the maximum AF in six pollutants is NO2 using 0 μg/m3 as the reference concentration which we estimated that about 32.68% (95% CI = 25.12% to 39.42%) corresponding to an AN of 5661 (95% CI = 4351 to 6827) RD-DM patients. In subgroup populations, we found similar results (Table S17 and Table S18 in the Online Supplementary Document). In terms of LOS and hospital cost, PM2.5, PM10, SO2, NO2 and O3 will significantly reduce more LOS and hospital cost using 0 μg/m3 as the reference, compared to using the standard of WHO’s 2021 AQG as the reference. And the AF of NO2 is highest. Gender-specific subgroup analyses found similar results (Table S19 and Table S21 in the Online Supplementary Document). However, age-specific subgroup analyses found different results (Table S20 and Table S22 in the Online Supplementary Document).

Table 2. The attributable risk for different population with different air quality guideline

WordPress Data Table

PM2.5 – particles with an aerodynamic diameter <2.5 micron (μm), PM10 – particles with an aerodynamic diameter <10 μm, SO2 – sulfur dioxide, NO2 – nitrogen dioxide, CO – carbon monoxide, LOS – length of hospital stay

The nonlinear relationship between air pollution and hospital admissions

Figure S10 in the Online Supplementary Document shows the exposure-response relationship for six air pollutants and hospital admissions. Each air pollutant exhibited approximately U-shaped curves with different degrees of curvature, indicating that air pollution affects hospital admissions in a nonlinear way. For example, the slopes are flat at low PM10 concentrations, while the slopes are steeper at 50 μg/m3 and higher concentrations.

Sensitivity analysis

The results of two-pollutant models at lag 07-day were showed in Table 3. After adjustment for other air pollutants, the effect estimates were still statistically significant. Table S14 to Table S16 in the Online Supplementary Document showed the results of two-pollutant models along different single-day lag structures. Figure S4 to Figure S6 in the Online Supplementary Document showed the results based on CHAP data were largely consistent with results based on fixed monitoring stations. Additionally, our results remained relatively stable after df of time was adjusted (4-12 df per year) (Figure S7 to Figure S9 in the Online Supplementary Document). And almost all of the E-values calculated for the total population along different lag structures exceed the original RR (Table S23 in the Online Supplementary Document).

Table 3. The effect of six air pollutants on admissions, length of hospital stay (LOS), and hospital cost for diabetes mellitus and comorbid respiratory diseases (RD-MD) associated with one unit increase in six air pollutants at lag 07 days using single- and two-pollutant models

WordPress Data Table

PM2.5 – particles with an aerodynamic diameter <2.5 micron (μm), PM10 – particles with an aerodynamic diameter <10 μm, SO2 – sulfur dioxide, NO2 – nitrogen dioxide, CO – carbon monoxide, O3 – ozone, LOS – length of hospital stay

*P < 0.05.

DISCUSSION

The relationship between air pollution and RD or DM has been estimated in previous studies. To our knowledge, this is one of the few studies to investigate the hospital admissions, LOS, and hospital cost among patients with RD-DM due to air pollution. Our results suggest that daily PM2.5, PM10, SO2, NO2, and CO contributed significantly to elevated risk of hospital admissions, LOS, and hospital cost for patients with RD-DM. Overall, 32.68% of hospital admissions for RD-DM could be attributed to NO2, with the highest population attributable faction. O3 showed protective effects on hospital admissions, LOS, and hospital cost among RD-DM patients, though two-pollutant models suggest the potential joint risk effects of O3 with other pollutants. The effects of air pollutants on hospital admissions, LOS and economic cost for RD-DM were more pronounced in people over the age of 65 and male patients than their counterparts.

The effects of most air pollutants on hospital admissions for RD-DM in lag 0 were relatively low and gradually reached a maximum effect at lag 07 in this study. A previous study focusing on only RD or DM suggested different trends. For example, a study of 234 279 patients hospitalised due to RD suggested that the effects of all pollutants except SO2 were highest in lag 0 while there was a gradual decrease in the cumulative effects of PM2.5 and NO2 [41]. A study of 88 904 patients with DM suggested an increased risk of hospital admissions for DM at lag 0-2 days [33]. Moreover, our result suggested that increasing PM2.5 by 10 mg/m3 could increase hospital admission risk for RD-DM in lag 2 by 3.38% (95% CI = 1.45 to 5.34%). A previous study suggested an increase of 10 ug/m3 in PM2.5 could increase the risk of hospital admissions for DM at lag 0-2 days by 1.71% (95% CI = 0.56 to 2.87%) [33], which is lower than the effect observed in our study. A meta-analysis also suggested that PM2.5 had a greater effect size in patients with RD-DM than those only with RD [42].

There are some possible explanations for the observed stronger effect of air pollutants on hospital admissions for patients with RD-DM. First, for individuals with comorbid DR and DM, disorders in both the respiratory and endocrine systems may grant them poorer immune responses and make them more susceptible to air pollutants. Previous studies have suggested that chronic hyperglycaemia can affect innate immune system acting on chemotaxis, phagocytosis and bactericidal activity of neutrophils and macrophages [43]. In animal models, inflammation, lung parenchyma, and vascular impairment are all associated with hyperglycaemia, and high levels of glucose in the airways can promote pathogen growth, decreased lung function [4,44]. Second, our study was conducted in a site with typical DHVs climate. The difference of climate conditions may affect the disease incidence [45]. Our study area has a low level of air pollution, e.g. the median value of SO2 (30.0 ug/m3) is lower than the criteria of WHO (40.0 ug/m3 for SO2), yet the strong effect of air pollutants on hospital admissions for RD-DM confirmed the previous conclusions that the health vulnerability of populations in areas with low air pollution exposure level is still noteworthy [46]. Third, heterogeneity in chemical components of air pollutants and demographic composition (e.g. a higher level of ageing in Panzhihua than the average level in China (15.9 vs. 14.9%) [47]) might also account for the different health effects [48].

Intriguingly, our study and sensitive analysis with other data revealed a negative association between O3 and hospital admissions for RD-DM, which is consistent with previous studies [49,50]. A potential explanation is that O3 is negatively associated with PM levels and thus the protective effect is merely an artifact of the negative association [49]. As our multiple pollutant models suggested, O3 combined with other pollutants, especially PM2.5, would increase the risk of hospital admissions for RD-DM. Another potential explanation is that the levels of O3 in our study site are lower than the Chinese National Ambient Air Quality Standard [51], which is in a range that has a protective effect on health. Since inconsistent evidence also exists [52], more studies are still needed to clarify the effect of O3.

Significant associations between air pollutants and LOS and hospital cost for RD-DM were observed in our study. There are some similarities and differences in the effects of specific pollutants compared to current studies. First, similar to Li Zhiwei’s study [19], our study suggested that PM will significantly increase LOS and hospital cost due to DM being comorbid with other diseases. A previous study in United States suggested that hospital cost increased by 47 dollars when monthly PM2.5 increased by 1 mg/m3 [53]. Second, our study suggested that the lag day of maximum effect value for LOS and hospital cost due to NO2 was 20.00 days (95% CI = 15.24 to 24.76) and 21 551.46 CNY (95% CI = 15 750.94 to 27 351.97) at lag 07. However, a previous study in China found the rise in 16-day average NO2 by 10-μg/m3 was correlated with negative alterations in LOS and hospital cost for patients with T2D [54]. Third, we found the LOS attributable fraction of NO2 is highest in all pollutants and subgroups, which is inconsistent to the previous study focusing on CVD-DM which suggests PM2.5 had the maximum attributable fraction [19]. As a common urban pollutant and precursor of PM and O3, our findings indicated that the effect of NO2 on LOS for respiratory system could be already notable before it alters to other chemical components.

We found some vulnerable subpopulation under the impacts of air pollution. Specifically, the effect of air pollution on participants >65 years old and male were slightly higher than their counterparts. This might be explained by poorer immune and metabolism function, and worse health condition in the elderly due to ageing and other comorbidities [12,55]. Besides, previous environmental epidemiological studies suggested that males are vulnerable to air pollution [13,35] (though exception exists [49]), due to males tend to have unhealthy lifestyles and spend more time outdoor than females, which may make them more likely to inhale pollutants [56]. For those who are frequently hospitalised due to poor condition, conceivably, their LOS and hospital cost would consequently increase, which might explain why we observe a similar trend of air pollutants’ effects on hospital admissions, LOS, and hospital cost in the total population and different subgroups.

Some potential mechanisms should be declared about how these pollutants affect the health of RD-DM patients. First, exposure to air pollutants may lead to the release of cytokines and inflammatory factors, which could further induce lung inflammation, reduce insulin sensitivity, and block the uptake of glucose in peripheral tissues [57]. Second, air pollutants can elicit inflammation that may lead to the risk of metabolic syndrome including diabetes, hypertension, and obesity [58]. Third, air pollutants may could affect the gastrointestinal tract, and may further affect the function of gut microbiota to affect the prognosis of DM [59].

Limitations

Some limitations should be mentioned. First, we fail to capture the dynamic movement of individuals with the assumption that all participants were with the same levels of ambient air pollutants exposure based on fixed monitoring stations, which may result in measurement errors. Second, the focus of our study was only on patients with total RD comorbid and total DM, and may not all subtypes of these two diseases are related to air pollution. Third, we fail to consider the interaction between variables since GAM has an inherent limitation in restricting the impact of various interactions. Fourth, the absence of some survey data, such as other environmental factors (e.g. noise, green space) and lifestyle habits (e.g. physical activity, passive smoking, diet), may limit the accuracy of the results. Further studies taking into account these potential confounders are needed to verify the relationship between air pollution and comorbidities. Fifth, we cannot draw causal inferences since this is an observational study and some important confounding factors may fail to consider. More studies that include the more information abovementioned can help strengthen our findings in the future. Sixth, due to limited ambient air pollution data availability, it was impossible to adjust all the ambient air pollution, such as Volatile Organic Compounds (VOCs), Nitrogen Oxides (NOx), dust and smoke.

CONCLUSIONS

This study found that most ambient air pollutants (PM2.5, PM10, SO2, NO2, and CO) were associated with increased risk of hospital admissions, LOS, and hospital cost for patients with RD-DM, and the effects were pronounced in people over the age of 65 and male patients. Our study complements the evidence on the relationship between ambient air pollutants and comorbidity from multi-dimensional measurement of risk, hospitalisation time, and economic cost in low-pollution areas of LMICs. Policy changes to reduce air pollutants exposure may lead to improved RD-DM admissions and substantial savings in health care spending and hospital stay.

Additional material

Online Supplementary Document

Acknowledgements

We are grateful to the data support from “National Earth System Science Data Center, National Science & Technology Infrastructure of China. (http://www.geodata.cn)”.

Disclaimer: The funders had no role in study design, data collection, data analysis, the decision to publication and preparation of the manuscript.

Ethics statement: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Ethics Committee of Panzhihua Central Hospital (Approval Number: pzhszxyyll-2021-63). Records of participants were anonymised before analysis.

Data availability: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

[1] Financial disclosure Funding: This work was supported by Scientific Research Project of Panzhihua Central Hospital (grant numbers 202102, 202201, 202301, 202302, 202315), Special Basic Cooperative Research Programs of Yunnan Provincial Undergraduate Universities’ Association (grant numbers 202101AO070272, 202101BA070001-220), Sichuan Science and Technology Program (grant numbers 21ZYZF-S-01), Research Programme of Panzhihua Science and Technology Bureau (grant number 2021ZX-5-1, 2022ZD-S-12, 2022ZD-S-24, 2022ZD-S-26) and Sichuan Province key clinical specialty construction Project, China.

[2] Authorship contributions: Li X: conceptualisation, data curation, formal analysis, funding acquisition, investigation, methodology, supervision, validation, visualisation, writing-original draft, writing-review & editing; Li Y and Yu B: supervision, validation, visualisation, writing-original draft, writing-review & editing; Meng H: data curation, supervision, validation; Shen M, Yang Y, Zhou Z, Liu S and Tian Y: writing-review & editing, Yin L and Xing X: supervision, validation, visualisation, writing-review & editing.

[3] Conflicts of interest Disclosure of interest: The authors completed the ICMJE Disclosure of Interest Form (available upon request from the corresponding author) and disclose no relevant interests.

references

1 

GBD 2017 Causes of Death Collaborators Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. 2018;392:1736-88. 10.1016/S0140-6736(18)32203-730496103

2 

S SafiriK Carson-ChahhoudM NooriSA NejadghaderiMJM SullmanJ Ahmadian HerisBurden of chronic obstructive pulmonary disease and its attributable risk factors in 204 countries and territories, 1990-2019: results from the Global Burden of Disease Study 2019. BMJ. 2022;378:e069679. 10.1136/bmj-2021-06967935896191

3 

GBD 2019 LRI Collaborators Age-sex differences in the global burden of lower respiratory infections and risk factors, 1990-2019: results from the Global Burden of Disease Study 2019. Lancet Infect Dis. 2022;22:1626-47. 10.1016/S1473-3099(22)00510-235964613

4 

D ViscaP PignattiA SpanevelloE LuciniE La RoccaRelationship between diabetes and respiratory diseases-Clinical and therapeutic aspects. Pharmacol Res. 2018;137:230-5. 10.1016/j.phrs.2018.10.00830312663

5 

S SafiriN KaramzadJS KaufmanAW BellSA NejadghaderiMJM SullmanPrevalence, Deaths and Disability-Adjusted-Life-Years (DALYs) Due to Type 2 Diabetes and Its Attributable Risk Factors in 204 Countries and Territories, 1990-2019: Results From the Global Burden of Disease Study 2019. Front Endocrinol (Lausanne). 2022;13:838027. 10.3389/fendo.2022.83802735282442

6 

B JeonFS LuysterSM SereikaMM DiNardoJA CallanER ChasensComorbid obstructive sleep apnea and insomnia and its associations with mood and diabetes-related distress in type 2 diabetes mellitus. Journal of clinical sleep medicine: JCSM: official publication of the American Academy of Sleep Medicine. 2022;18:1103-11. 10.5664/jcsm.981234879902

7 

R KaurT EgliJ PaynterR MurphyL PerumalA LeeTuberculosis and diabetes: increased hospitalisations and mortality associated with renal impairment. Intern Med J. 2023;53:1588-94. 10.1111/imj.1566834936168

8 

RN AuroraNM PunjabiObstructive sleep apnoea and type 2 diabetes mellitus: a bidirectional association. Lancet Respir Med. 2013;1:329-38. 10.1016/S2213-2600(13)70039-024429158

9 

P ZhangR LopezAH AttawaySN GeorasSB KhatriS Abi-SalehDiabetes Mellitus Is Associated with Worse Outcome in Patients Hospitalized for Asthma. J Allergy Clin Immunol Pract. 2021;9:1562-9.e1. 10.1016/j.jaip.2020.10.05433181340

10 

GBD 2019 Risk Factors CollaboratorsGlobal burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396:1223-49. 10.1016/S0140-6736(20)30752-233069327

11 

GC KhilnaniP TiwariAir pollution in India and related adverse respiratory health effects: past, present, and future directions. Curr Opin Pulm Med. 2018;24:108-16. 10.1097/MCP.000000000000046329300211

12 

A SalminenActivation of immunosuppressive network in the aging process. Ageing Res Rev. 2020;57:100998. 10.1016/j.arr.2019.10099831838128

13 

A ShanY ZhangLW ZhangX ChenX LiH WuAssociations between the incidence and mortality rates of type 2 diabetes mellitus and long-term exposure to ambient air pollution: A 12-year cohort study in northern China. Environ Res. 2020;186:109551. 10.1016/j.envres.2020.10955132330771

14 

AA AlmetwallyM Bin-JumahAA AllamAmbient air pollution and its influence on human health and welfare: an overview. Environ Sci Pollut Res Int. 2020;27:24815-30. 10.1007/s11356-020-09042-232363462

15 

BY YangS FanE ThieringJ SeisslerD NowakGH DongAmbient air pollution and diabetes: A systematic review and meta-analysis. Environ Res. 2020;180:108817. 10.1016/j.envres.2019.10881731627156

16 

IC EzeLG HemkensHC BucherB HoffmannC SchindlerN KünzliAssociation between ambient air pollution and diabetes mellitus in Europe and North America: systematic review and meta-analysis. Environ Health Perspect. 2015;123:381-9. 10.1289/ehp.130782325625876

17 

H ZhangY NiuY YaoR ChenX ZhouH KanThe Impact of Ambient Air Pollution on Daily Hospital Visits for Various Respiratory Diseases and the Relevant Medical Expenditures in Shanghai, China. Int J Environ Res Public Health. 2018;15:425. 10.3390/ijerph1503042529495633

18 

A PhosriK UedaVLH PhungB TawatsupaA HondaH TakanoEffects of ambient air pollution on daily hospital admissions for respiratory and cardiovascular diseases in Bangkok, Thailand. Sci Total Environ. 2019;651:1144-53. 10.1016/j.scitotenv.2018.09.18330360246

19 

Z LiM LiuZ WuY LiuW LiM LiuAssociation between ambient air pollution and hospital admissions, length of hospital stay and hospital cost for patients with cardiovascular diseases and comorbid diabetes mellitus: Base on 1,969,755 cases in Beijing, China, 2014-2019. Environ Int. 2022;165:107301. 10.1016/j.envint.2022.10730135598418

20 

Y ZhangSG WangYX MaKZ ShangYF ChengX LiAssociation between Ambient Air Pollution and Hospital Emergency Admissions for Respiratory and Cardiovascular Diseases in Beijing: a Time Series Study. Biomed Environ Sci. 2015;28:352-63.26055562

21 

K AdamsDS GreenbaumR ShaikhAM van ErpAG RussellParticulate matter components, sources, and health: Systematic approaches to testing effects. J Air Waste Manag Assoc. 2015;65:544-58. 10.1080/10962247.2014.100188425947313

22 

R ChenRD PengX MengZ ZhouB ChenH KanSeasonal variation in the acute effect of particulate air pollution on mortality in the China Air Pollution and Health Effects Study (CAPES). Sci Total Environ. 2013;450-451:259-65. 10.1016/j.scitotenv.2013.02.04023500824

23 

Y ZhangZ DingQ XiangW WangL HuangF MaoShort-term effects of ambient PM(1) and PM(2.5) air pollution on hospital admission for respiratory diseases: Case-crossover evidence from Shenzhen, China. Int J Hyg Environ Health. 2020;224:113418. 10.1016/j.ijheh.2019.11.00131753527

24 

X LiY LiB YuH ZhuZ ZhouY YangHealth and economic impacts of ambient air pollution on hospital admissions for overall and specific cardiovascular diseases in Panzhihua, Southwestern China. J Glob Health. 2022;12:11012. 10.7189/jogh.12.1101236538381

25 

B HoffmannH BoogaardA de NazelleZJ AndersenM AbramsonM BrauerWHO Air Quality Guidelines 2021-Aiming for Healthier Air for all: A Joint Statement by Medical, Public Health, Scientific Societies and Patient Representative Organisations. Int J Public Health. 2021;66:1604465. 10.3389/ijph.2021.160446534630006

26 

H QiuVC PunL TianShort-term effects of fine and coarse particles on deaths in Hong Kong elderly population: An analysis of mortality displacement. Environmental pollution (Barking, Essex: 1987). 2018;241:148-54. 10.1016/j.envpol.2018.05.05629804047

27 

H QiuL WangL ZhouJ PanCoarse particles (PM(2.5-10)) and cause-specific hospitalizations in southwestern China: Association, attributable risk and economic costs. Environ Res. 2020;190:110004. 10.1016/j.envres.2020.11000432745536

28 

S SunF LadenJE HartH QiuY WangCM WongSeasonal temperature variability and emergency hospital admissions for respiratory diseases: a population-based cohort study. Thorax. 2018;73:951-8. 10.1136/thoraxjnl-2017-21133329622691

29 

D CaoD ZhengZM QianH ShenY LiuQ LiuAmbient sulfur dioxide and hospital expenditures and length of hospital stay for respiratory diseases: A multicity study in China. Ecotoxicol Environ Saf. 2022;229:113082. 10.1016/j.ecoenv.2021.11308234929503

30 

C BurnsG SacksL GoldLongitudinal study of Consumer Price Index (CPI) trends in core and non-core foods in Australia. Aust N Z J Public Health. 2008;32:450-3. 10.1111/j.1753-6405.2008.00278.x18959549

31 

MR ZonfrilloI ZanilettiM HallES FieldstonJD ColvinJL BettenhausenSocioeconomic Status and Hospitalization Costs for Children with Brain and Spinal Cord Injury. J Pediatr. 2016;169:250-5. 10.1016/j.jpeds.2015.10.04326563534

32 

J SongM LuJ LuL ChaoZ AnY LiuAcute effect of ambient air pollution on hospitalization in patients with hypertension: A time-series study in Shijiazhuang, China. Ecotoxicol Environ Saf. 2019;170:286-92. 10.1016/j.ecoenv.2018.11.12530530180

33 

H LuoC LiuX ChenJ LeiY ZhuL ZhouAmbient air pollution and hospitalization for type 2 diabetes in China: A nationwide, individual-level case-crossover study. Environ Res. 2023;216:114596. 10.1016/j.envres.2022.11459636272593

34 

W LiuJ WeiM CaiZ QianZ LongL WangParticulate matter pollution and asthma mortality in China: A nationwide time-stratified case-crossover study from 2015 to 2020. Chemosphere. 2022;308:136316. 10.1016/j.chemosphere.2022.13631636084833

35 

H LuoQ ZhangK YuX MengH KanR ChenLong-term exposure to ambient air pollution is a risk factor for trajectory of cardiometabolic multimorbidity: A prospective study in the UK Biobank. EBioMedicine. 2022;84:104282. 10.1016/j.ebiom.2022.10428236174399

36 

D CaoD LiY WuZM QianY LiuQ LiuAmbient PM(2.5) exposure and hospital cost and length of hospital stay for respiratory diseases in 11 cities in Shanxi Province, China. Thorax. 2021. 10.1136/thoraxjnl-2020-21583834088786

37 

World Health Organization. WHO global air quality guidelines: Particulate matter (PM(25) and PM(10)), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. Geneva: World Health Organization;2021.

38 

K TakeleT ZewotirD NdanguzaUnderstanding correlates of child stunting in Ethiopia using generalized linear mixed models. BMC Public Health. 2019;19:626. 10.1186/s12889-019-6984-x31118013

39 

A ZekaA ZanobettiJ SchwartzIndividual-level modifiers of the effects of particulate matter on daily mortality. Am J Epidemiol. 2006;163:849-59. 10.1093/aje/kwj11616554348

40 

TJ VanderWeeleP DingSensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167:268-74. 10.7326/M16-260728693043

41 

H JiaJ XuL NingT FengP CaoS GaoAmbient air pollution, temperature and hospital admissions due to respiratory diseases in a cold, industrial city. J Glob Health. 2022;12:04085. 10.7189/jogh.12.0408536243957

42 

M LiuZ LiF LuM GuoL TaoM LiuAcute effect of particulate matter pollution on hospital admissions for cause-specific respiratory diseases among patients with and without type 2 diabetes in Beijing, China, from 2014 to 2020. Ecotoxicol Environ Saf. 2021;226:112794. 10.1016/j.ecoenv.2021.11279434592518

43 

NH ValeriusC EffNE HansenH KarleJ NerupB SøebergNeutrophil and lymphocyte function in patients with diabetes mellitus. Acta Med Scand. 1982;211:463-7. 10.1111/j.0954-6820.1982.tb01983.x6981286

44 

BJ PhilipsJ RedmanA BrennanD WoodR HollimanD BainesGlucose in bronchial aspirates increases the risk of respiratory MRSA in intubated patients. Thorax. 2005;60:761-4. 10.1136/thx.2004.03576616135681

45 

PL KinneyInteractions of Climate Change, Air Pollution, and Human Health. Curr Environ Health Rep. 2018;5:179-86. 10.1007/s40572-018-0188-x29417451

46 

JO KlompmakerJE HartP JamesMB SabathX WuA ZanobettiAir pollution and cardiovascular disease hospitalization – Are associations modified by greenness, temperature and humidity? Environ Int. 2021;156:106715. 10.1016/j.envint.2021.10671534218186

47 

National Bureau of Statistics. Statistical Bulletin of the People’s Republic of China on National Economic and Social Development in 2022. Available: http://www.stats.gov.cn/sj/zxfb/202302/t20230228_1919011.html. Accessed: 2 September 2023.

48 

D GaoS RipleyS WeichenthalKJ Godri PollittAmbient particulate matter oxidative potential: Chemical determinants, associated health effects, and strategies for risk management. Free Radic Biol Med. 2020;151:7-25. 10.1016/j.freeradbiomed.2020.04.02832430137

49 

H LiD DuanJ XuX FengT Astell-BurtT HeAmbient air pollution and risk of type 2 diabetes in the Chinese. Environ Sci Pollut Res Int. 2019;26:16261-73. 10.1007/s11356-019-04971-z30977004

50 

AD BergstraJV BeenA BurdorfThe association of specific industry-related air pollution with occurrence of chronic diseases: A register-based study. Environ Res. 2022;209:112872. 10.1016/j.envres.2022.11287235131328

51 

Z WangY TanM GuoM ChengY GuS ChenProspect of China’s ambient air quality standards. J Environ Sci (China). 2023;123:255-69. 10.1016/j.jes.2022.03.03636521988

52 

YL LiTW ChuangPY ChangLY LinCT SuLN ChienLong-term exposure to ozone and sulfur dioxide increases the incidence of type 2 diabetes mellitus among aged 30 to 50 adult population. Environ Res. 2021;194:110624. 10.1016/j.envres.2020.11062433412098

53 

N FannSY KimC OlivesL SheppardEstimated Changes in Life Expectancy and Adult Mortality Resulting from Declining PM2.5 Exposures in the Contiguous United States: 1980-2010. Environ Health Perspect. 2017;125:097003. 10.1289/EHP50728934094

54 

X LiK TangXR JinY XiangJ XuLL YangShort-term air pollution exposure is associated with hospital length of stay and hospitalization costs among inpatients with type 2 diabetes: a hospital-based study. J Toxicol Environ Health A. 2018;81:819-29. 10.1080/15287394.2018.149191230015599

55 

A Singh-ManouxA FayosseS SabiaA TabakM ShipleyA DugravotClinical, socioeconomic, and behavioural factors at age 50 years and risk of cardiometabolic multimorbidity and mortality: A cohort study. PLoS Med. 2018;15:e1002571. 10.1371/journal.pmed.100257129782486

56 

C VlassoffGender differences in determinants and consequences of health and illness. J Health Popul Nutr. 2007;25:47-61.17615903

57 

Y LiL XuZ ShanW TengC HanAssociation between air pollution and type 2 diabetes: an updated review of the literature. Ther Adv Endocrinol Metab. 2019;10:2042018819897046. 10.1177/204201881989704631903180

58 

EA ClementiA TalusanS VaidyanathanA VeerappanM MikhailD OstrofskyMetabolic Syndrome and Air Pollution: A Narrative Review of Their Cardiopulmonary Effects. Toxics. 2019;7:6. 10.3390/toxics701000630704059

59 

F FouladiMJ BaileyWB PattersonM SiodaIC BlakleyAA FodorAir pollution exposure is associated with the gut microbiome as revealed by shotgun metagenomic sequencing. Environ Int. 2020;138:105604. 10.1016/j.envint.2020.10560432135388

Correspondence to:
Xiangyi Xing
Panzhihua Central Hospital
No.34, Yikang Street, East District, Panzhihua
China
[email protected]
Li Yin
Panzhihua Central Hospital
No.34, Yikang Street, East District, Panzhihua
China
[email protected]