Short-term effects of ambient air pollution on daily emergency room visits for abdominal pain: a time-series study in Wuhan, China

Short-term exposure to ambient air pollution has been proven to result in respiratory, cardiovascular, and digestive diseases, leading to increased emergency room visits (ERVs). Abdominal pain complaints provide a large proportion of the ERVs, as yet few studies have focused on the correlations between ambient air pollution and abdominal pain, especially in emergency departments within China. Daily data for daily ERVs were collected in Wuhan, China (from January 1, 2016 to December 31, 2018), including air pollution concentration (SO2, NO2, PM2.5, PM10, CO, and O3), and meteorological variables. We conducted a time-series study to investigate the potential correlation between six ambient air pollutants and ERVs for abdominal pain and their effects, in different genders, ages, and seasons. A total of 16,318 abdominal pain ERVs were identified during the study period. A 10-μg/m3 increase in concentration of SO2, NO2, PM2.5, PM10, CO, and O3 corresponded respectively to incremental increases in abdominal pain of 4.89% (95% confidence interval [CI]: − 1.50–11.70), 1.85% (95% CI: − 0.29–4.03), 0.83% (95% CI: − 0.05–1.72), − 0.22% (95% CI: − 0.73–0.30), 0.24% (95% CI: 0.08–0.40), and 0.86% (95% CI: 0.04 − 1.69). We observed significant correlations between CO and O3 and increases in daily abdominal pain ERVs and positive but insignificant correlations between the other pollutants and ERVs (except PM10). The effects were stronger for females (especially SO2 and O3: 13.53% vs. − 2.46%; 1.20% vs. 0.47%, respectively) and younger people (especially CO and O3: 0.25% vs. 0.01%; 1.36% vs. 0.15%, respectively). Males (1.38% vs. 0.87%) and elders (1.27% vs. 0.99%) were more likely to be affected by PM2.5. The correlations with PM2.5 were stronger in cool seasons (1.25% vs. − 0.07%) while the correlation with CO was stronger in warm seasons (0.47% vs. 0.14%). Our time-series study suggests that short-term exposure to air pollution (especially CO and O3) was positively correlated with ERVs for abdominal pain in Wuhan, China, and that the effects varied by season, gender and age. These data can add evidence on how air pollutants affect the human body and may prompt hospitals to take specific precautions on polluted days and maintain order in emergency departments made busier due to the pollution.


Introduction
The Global Burden of Disease study identified air pollution as a leading cause of global disease burden, especially in developing countries (Collaborators 2018). Short-term exposure to ambient air pollution can acutely exacerbate respiratory, cardiovascular, and digestive diseases, leading to increases in emergency room visits (ERVs) (Rodopoulou et al. 2015, Vignal et al. 2021. Since Wuhan is one of the twenty most polluted cities in China, with a large number of daily ERVs, and since China is the largest developing country in the world, it is important to figure out potential health risks from ambient air pollution.
Recently, there has been an accumulation of articles revealing the effects of air pollution in hospital emergency departments (Rodopoulou et al. 2015;Chen et al. 2019), which plays a significant role in dealing with public health events. However, untimely treatment caused by overcrowding of emergency departments is an international issue, among which abdominal pain (ICD-10: R10.4) is the most significant proportion (Hooker et al. 2019). The most common diseases manifesting abdominal pain in ERVs are gastroenteritis (ICD-10: K52.9), cholecystitis (ICD-10: K81.0), and urolithiasis (ICD-10: N20.9), and these will cause acute and severe pain, and thus need emergency treatments. Environmental pollution has been shown to play an important role in the development of these diseases. For instance, exposure to the particulate matter may lead to an increase in urolithiasis, by causing urine volume decreases through vascular endothelial injury, systemic inflammation, atherosclerosis, and microvascular changes (Sun et al. 2005, Chow et al. 2006. Air pollution exposure can increase the hospitalizations for digestive diseases, including inflammatory bowel disease, peptic ulcers, and enteritis (Ananthakrishnan et al. 2011, Xu et al. 2016, Tsai et al. 2019, Gu et al. 2020, although the mechanisms remain unclear and inconclusive. However, studies which focus on the pollution effects on specific diseases or symptoms in ERVs, such as these abdominal diseases, are still lacking. It will be beneficial for hospitals to figure out the correlations between air pollution and abdominal pain, and thereby to understand the characteristics of ERVs, take precautions, and maintain order on particularly heavily polluted days. Considering the above-mentioned reports, we conducted this time-series study to analyze the correlations between six ambient air pollutants (SO 2 , NO 2 , PM 2.5 , PM 10 , CO, O 3 ) and ERVs for abdominal pain, in Zhongnan Hospital of Wuhan University from 1 January 2016 to 31 December 2018. We also analyzed the discrepancy correlation between different seasons, gender, and age. In addition, we explored the exposure-response relationship curves between ERVs for abdominal pain and air pollutants, as well as the co-effects among six air pollutants. Our results should be useful in risk assessment, resource allocation, and health policy-making.

Materials and methods
Wuhan is the capital of Hubei Province, with a population size over 10 million, and lying in Central China (latitude 30°35'N and longitude 114°17'E), and covering 8494.41 km 2 . It consists of seven central districts and six suburban and rural districts (Fig. 1). Because it is an essential industrial and transport interchange of China, the main causes of air pollution are vehicle exhaust and industrial emissions. The northerly wind transports polluted air from heavily polluted provinces from the north, such as Henan Province and Hebei Province, exerting a profound impact on air pollution of Wuhan as well (Mao et al. 2018). Wuhan has a humid monsoon subtropical climate. Summers are hot and wet while winters are cold and dry. The average temperature is 30.3 °C in July and 4.4 °C in January.

Emergency room visit data
The majority of the ERVs of Zhongnan Hospital are from the Wuchang District of Wuhan, where the hospital is located. It is a district which contains 19% population of the whole urban area and 12% population of Wuhan. We were provided with the daily number of ERVs from January 1, 2016 to December 31, 2018, from Zhongnan Hospital of Wuhan University in Wuhan, China. The ERV data followed a quasi-Poisson distribution. From the ERV records, we specifically selected those outpatient visits whose chief complaint was abdominal pain (N = 16,318). The research protocol was approved by the Medical Ethics Committee of Zhongnan Hospital (IRB number: 2021018 K).

Environmental and meteorological data
We acquired daily ambient air pollution data (SO 2 , NO 2 , PM 2.5 , PM 10 , CO, and O 3 ) from January 2016 to December 2018, from the website of the Wuhan Ecological Environment Bureau (http:// hbj. wuhan. gov. cn/). The daily average concentrations of air pollutant were calculated by average hourly values from ten fixed-site stationary centers, which can cover the urban districts of Wuhan. All the stations are located away from industrial, residential, and vehicle sources, to make sure they can monitor the representative measurements of background pollution, without undue interference.
The data of two meteorological parameters during the study period (average daily temperature [℃] and relative humidity [%]) were acquired from the Meteorological Data Sharing Service System of the China Meteorological Administration (Beijing, China). Thirty-two days (2.92%) of environmental and meteorological data were missing. Dates with missing data were eliminated from our study.

Statistical analysis
We chose the over-dispersed generalized additive model (GAM) to perform a time-series analysis and to explore the acute effect of air pollutants on ERVs with abdominal pain while controlling for both time-invariant and time-varying confounders. This model can more precisely result in non-linear relationship data sets automatically, and its discussion of "lag" is quite important in air quality-related studies (Ravindra et al. 2019). Some covariates were added into the main model. First, a natural cubic regression smoothing function of calendar time with 7 degrees of freedom (df) per year excluded unmeasured long-term and seasonal trends longer than 2 months (Zhou et al. 2021). Second, a natural smooth function of the average daily temperature (6 df) and relative humidity (3 df) controlled for the nonlinear confounding effects of weather conditions (Song et al. 2019). Third, other covariates such as day of the week (DOW) and public holidays (Holiday) were controlled as dummy variables in the basic models.
The core model can be described as follows: where E (Yt) represents the number of abdominal pain count for day t; β represents the log-related rate of EVRs with abdominal pain associated with a unit increase of air pollutants; Zt refers to the pollutant concentrations at day t; DOW is a dummy variable for day of the week; and ns (time, df) means the natural cubic regression smooth function. We plotted the exposure-response relationship curves between air pollutant and ERVs with abdominal pain by adding to the above model a spline function with 3 df.
We conducted 3 sensitivity analyses to ensure the stability of our models. First, Two-pollutant models can examine the robustness of effect estimates. For this, co-pollutants with a correlation coefficient < 0.7 would be added to the model. Second, we conducted two different lag constructions: singleday lags (lag0 to lag7), and moving average lags (lag 01 to lag 07). We chose the best lag structure according to the Akaike information criterion (AIC) and the generalized cross validation (GCV) of the models. Third, we selected alternative df with 4-10 per year. The Q-Q plots of deviance residuals, the response vs. fitted values, the histogram of residuals, and the residuals vs. linear predictor were drawn to verify the fitting effect of model (Fig. S1).
To investigate the discrepant correlations among different groups of people, three stratification analyses were conducted according to season (cool: October to March; warm: April to September), sex, and age (< 45 years old and ≥ 45 years old).
All statistical models were run in R software (version 3.6.0) using the MGCV package. The statistical tests were two-sided, with correlations with p < 0.05 mean statistically significant. The effects are described as the percent changes and 95% CI in daily ERVs for abdominal pain per 10 μg/m 3 increase of each pollutant.
The exposure-response (E-R) curves for the correlations between pollutants at their peak lag day and abdominal pain ERVs were generally positive (Fig. 5). The E-R curve of SO 2 was an ascent after a slight decline at the concentration ≤ 8 μg/m 3 .The E-R curve of PM 2.5 showed a slowly growing slope. The E-R curves of NO 2 and CO increased sharply before the concentrations ≥ 45 μg/m 3 and ≥ 800 μg/ m 3 , respectively and then showed just a slight rise. The E-R of PM 10 increased slowly at the concentration ≤ 60 μg/m 3 then turned into a slight decline. The E-R curve of O 3 was a rising slope at the concentration ≤ 130 μg/m 3 then became flat. We did not observe evident threshold concentrations above which NO 2 , PM 2.5 , and CO were not correlated with ERVs. Table 3 summarizes the results for possible effects modified by season, gender, and age. It shows that PM 2.5 was significantly related to abdominal pain ERVs in cool seasons (1.25% vs. − 0.07%), while CO was significantly related in warm seasons (0.47% vs. 0.14%). The influence of SO 2 and O 3 were more obvious on females than on males (13.53% vs. − 2.46%; 1.20% vs. 0.47%, respectively) while PM 2.5 showed more significant effect on males (1.38% vs. 0.87%). For CO and O 3 , the correlations were significantly stronger among younger people (< 45 years old) than elders (0.25% vs. 0.01%; 1.36% vs. 0.15%, respectively), while elders were more likely to be affected by PM 2.5 (1.27% vs. 0.99%).
Pollutants with < 0.7 Spearman correlation co-efficient were added into two-pollutant models, and most models remained stable after adjusting for co-pollutants.
Especially for CO, its positive relation with daily ERVs for abdominal pain remained significantly robust after adjustment (Table 2). Another sensitivity analysis proved that alternative df (4-10) did not significantly affect the effects of pollutants on abdominal pain ERVs (Fig. S2).

Discussion
Abdominal pain is one of the most common complaints during ERVs (Hooker et al. 2019), and few studies have explored the acute effect of ambient air pollution on abdominal pain, especially in emergency room visits. Overall, our study suggests that short-term exposure to air pollutants, including CO and O 3 , is significantly correlated with increased risks of abdominal pain ERVs. The correlations between CO and abdominal pain ERVs were robust after adjustment in two-pollutant models. The correlations were stronger mainly on females (especially from SO 2 and O 3 ) and younger people (especially from CO and O 3 ), while the correlation of PM 2.5 was more obvious on males and elders. The correlation of PM 2.5 was stronger in cool seasons, but for CO, stronger in warm seasons. Our findings have added evidence on how air pollutants affect the human body and may help hospitals take precautions on polluted days and maintain order in emergency departments made busier due to the pollution.
During our study period, the annual mean concentrations of NO 2 (47.87 μg/m 3 ), PM 2.5 (50.79 μg/m 3 ), and PM 10 (83.55 μg/m 3 ) in Wuhan exceeded the Chinese National Ambient Air Quality Standards (40, 35, and 70 μg/m 3 ). Urban expansion, industrial development, the increase of motor vehicles, and an increasing population are major contributors to the effects of severe pollution. However, it is promising that the overall trend of air quality is improving.
Our study proved that short-term exposure of ambient air pollutants (SO 2 , NO 2 , PM 2.5 , CO, O 3 ) has positive correlations with the ERVs for abdominal pain, especially CO and O 3 . PM 10 showed a negative correlation with the ERVs for abdominal pain, but it was insignificant. Highly elevated or protracted CO exposures may produce untoward biological oxidations and interfere with homeostasis (Piantadosi 2008). CO can be readily absorbed from the lungs into the bloodstream, where it forms carboxyhemoglobin with hemoglobin, thus resulting in tissue hypoxia. The relationships between lipid peroxidation and membrane fatty acid composition in CO intoxication may also be a potential mechanism (Akyol et al. 2014). The cooperation of CO and PM 2.5 may also enhance the damage that was also shown in our study (the Spearman correlation co-efficient is > 0.7) but this is still not certain (Chen et al. 2007). For O 3 , the toxicity was due mainly to its action as an oxidant, causing lipid peroxide decomposition, and inflammatory and blood rheology (Mehlman and Borek 1987;Yang et al. 2019). The systemic response may also affect the digestive and urinary systems and lead to abdominal pain. Although O 3 has been regarded as a kind of medical drug gradually, we still suggest that more attention be given to its effect from different concentrations and the particular biological system where it acts (Zanardi et al. 2016).
The positive results varied by season, sex, and age. In our study, CO demonstrated a stronger effect in warm seasons than in cool seasons. We could not find a sufficiently convincing explanation for this in the literature, but tentatively, we attribute this to the differences in air pollution mixture between warm and cool seasons (Tsai et al. 2019). Thus, we suggest that the mechanism regarding how the mixed effects of air pollutants cooperating with weather condition that impacts people's health should be investigated further. PM 2.5 demonstrated higher effects during cool seasons in our study, which was consistent with previous studies (Zhang and Cao 2015;Song et al. 2018). This might be explained by the special geographic location of Wuhan and the unfavorable meteorological conditions in winter, which limit the dispersion of air pollutants and transport relatively dirty air from more polluted provinces (Mao et al. 2018).
Our study indicates that PM 2.5 would significantly increase the risk of abdominal pain of males and elders, which is consistent with previous studies (Kaplan et al. 2009, Tian et al. 2017, Gu et al. 2020). This may be due to more outdoor activities of males and cumulative toxic effects from long-term exposure to ambient pollution and more comorbidities of elders. We found also that females were more likely to suffer the effects of increases in SO 2 and O 3 . This phenomenon has been shown in the respiratory and cardiovascular systems as well, which was attributed to the greater airway reactivity, as well as smaller airways of females (Kan et al. 2008;Zhong et al. 2018). Most previous studies, showing ambient pollutants having more serious effect on elders, are contrary to our own earlier results, Fig. 5 The exposure-response relationship curves of SO 2 (lag01), NO 2 (lag05), PM 2.5 (lag6), PM 10 (lag2), CO (lag06), and O 3 (lag01) which show that CO and O 3 affected younger people more (Yazdi et al. 2021). We supposed these may be because of the acute effects on elders' center mainly on the cardiovascular and respiratory systems, rather than the digestive or urinary systems (Chen et al. 2007, Nuvolone et al. 2018, Lee et al. 2020 (Fig. 6). However, excessive focus on the effects of a single pollutant may ignore the cooperation between pollutants. Population-based observational studies should consider the correlations between them (Chen et al. 2007). The temporal-spatial risk analysis of air pollution is also needed, since individuals living in areas of chronically higher air pollution exposures are more likely to be affected by temporal air pollution rises (Liu et al. 2020).
The most common acute diseases in ERVs for abdominal pain are gastroenteritis, cholecystitis, and urolithiasis. The correlations between gastrointestinal diseases and air pollution have been widely discussed in recent years. Studies showed that air pollution exposure may increase the risk of appendicitis, as well as hospitalizations for inflammatory bowel disease, peptic ulcers, and enteritis (Kaplan et al. 2009, Ananthakrishnan et al. 2011, Xu et al. 2016, Tsai et al. 2019. A study in Canada found that air pollution may cause non-specific abdominal pain, especially in young ladies, which is consistent with our studies (Kaplan et al. 2012). The main mechanisms for pollutants increasing gastrointestinal diseases include directly toxic effects, systemic inflammation, immune activation, gut microbiota effects, and gut permeability effects (Beamish et al. 2011, Mutlu et al. 2011. But there were also contradictory studies that revealed no association whatever between air pollution and some gastrointestinal diseases (Kaplan et al. 2010;Quan et al. 2015). Different degrees of vulnerability and susceptibility between specific clusters (age, gender, ethnicity, economic status, and so on) of people and the various effects on different diseases may lead to the discrepant results (Kaplan et al. 2013). Studies about these correlations and results are greatly needed.
Cholecystitis caused by gallstone is one of the most common diseases in emergency department visits. There are at present no studies investigating the correlation between ambient air pollution and cholecystitis. Actually, cholecystic diseases have a close relationship with liver diseases but occur more commonly in ERVs. Studies have revealed that the obstruction of cholecystitis may result in hepatic inflammatory changes and cause liver diseases to progress to secondary biliary cirrhosis (Flinn et al. 1977;Geraghty and Goldin 1994). Chronic liver diseases will also impair the gallbladder contractility and contribute in return to an increased gallstone formation (Acalovschi et al. 2004). Long-term exposure to environmental pollutants may lead to liver abnormality or injury by activating reactive oxygen species (ROS) production and initial immune response (Sanchez-Valle et al. 2012;Kim et al. 2014), both of which have a chronic effect on the liver. As for acute effect, it is our supposition that ambient air pollution might also have a positive correlation with cholecystitis, through causing dyslipidemia (Zhang et al. 2021), which is considered as a risk factor of gallstones and cholecystitis (Lammert et al. 2016). Whether the liver injuries or the cholecystic diseases are the initiating agents for acute hepatobiliary diseases is still uncertain.
Furthermore, air pollution may increase the risk of bladder, kidney, and urinary tract cancer, as well as other more benign diseases (Al-Aly and Bowe 2020, Zare Sakhvidi et al. 2020). In an emergency department, the most common visits for abdominal pain in the urinary system are for urolithiasis. The effects of particulate matter can lead to vascular endothelial injury, systemic inflammation, atherosclerosis, Fig. 6 The potential factor that may cause the discrepant result and microvascular changes (Sun et al. 2005, Chow et al. 2006, and thus may cause urine volume decreases, which can result in urolithiasis. However, there are as yet no studies examining the correlation between urolithiasis and ambient air pollution, which bears further investigation. Our study has limitations. First, as in most previous studies, we used fixed-site monitor measurements to represent personal exposure, causing exposure errors. However, the resultant nondifferential error was reported to produce an underestimate of the correlations of ambient air pollution (Samet et al. 2000). Second, due to data limitations, we didn't take pre-existing diseases and unhealthy factors into account, either of which may weaken one's tolerance of air pollutants. Third, since our study was conducted only from emergency department data, some visits could not produce a definite diagnosis or could even get a wrong diagnosis, which is hard to avoid. Fourth, this study collected data from only one highly polluted city; thus, its general application might be limited.

Conclusions
Our time-series study suggests that ambient air pollution (especially CO and O 3 ) has a positive correlation with the ERVs for abdominal pain in Wuhan, China. The effects varied by season, gender, and age. The estimates of effects from most pollutants on ERVs for abdominal pain were mostly higher among females and younger people, except for PM 2.5 , which showed significantly higher effects on males and elders. The correlations between PM 2.5 ERVs for abdominal pain was stronger in cool seasons, while the correlation of CO was higher in warm seasons. Specifically, we focused more attention on the correlations between ambient air pollution and specific emergency diseases manifested as abdominal pain, and we expect that this study can help hospitals to take precaution on polluted days and improve timely treatment in hospitals.