Air Pollution Associated Dry Eye Disease May Be Increased by Qi Stagnation in Young: A Nested Case-Control Study in Liaoning, China

Background: Little information exists regarding the interaction of Qi stagnation and air pollution on young’s dry eye disease (DED) in xed area with an epidemic threat. The aim of this study is to assess the modication by Qi stagnation on the associations of exposure with DED in Chinese young. Methods: A nested case-control study was conducted in Chinese young population aged 17-20 years, within a cohort of He Eye Health screening project. All cases newly diagnosed (n = 576) were matched to controls (n = 576) by age and sex. Logistic regression model was used to assess the odds ratio (OR) and 95% condence interval (CI) of DED for exposure to different air pollutants in terms of an interquartile range (IQR) increase in exposure level, adjusted for potential confounders. Results: In single-pollutant model, the DED ORs (95% CI) for the 20 th to 40 th percent of quintile (vs. lowest) were 2.64 (1.66, 4.18) for PM 2.5 , 1.70 (1.09, 2.66) for PM 10 , for the highest quintile (vs. lowest) were 1.74 (1.13, 2.66) for CO, and 1.57 (0.98, 2.51) NO 2 . In contrast, the odds ratios for the highest quintile (vs. lowest) were 0.23 (0.07, 0.78) for O 3 . Corresponding estimates were higher among Stagnant qi constitution [OR PM10 = 1.69 (1.08, 2.16), OR CO = 1.93 (1.19, 3.54)] than Balanced constitution [OR PM10 = 1.29 (0.79, 2.05), OR CO = 1.70 (1.09, 3.43)] for PM 10 and CO 2 , respectively. In addition, O 3 was inversely associated for Stagnant qi constitution [0.67 (0.23, 1.90)] and Balanced constitution [0.51 (0.08, 1.40)]. Conclusion: Exposure to ambient PM 2.5 , PM 10 , CO and NO 2 was positively associated with


Introduction
Currently, many people in China suffer from serious dust-haze pollution that detracts from their ability to live in full health [1]. The government has implemented increasingly special policy to reduce air pollution, information exists regarding the interaction of physical environment (air pollution) and social environment (Qi) on DED.
Air pollution is main physical environment factor. An ecologic study suggested that short-term exposure of airborne particulate matter (PM), especially that less than 2.5 μm in diameter (PM 2.5 ), induced acute ocular surface diseases by altered activation of autophagy in a time-dependent manner [10]. Meanwhile, a cross-sectional study in China suggested that nitrogen dioxide (NO 2 ), high-temperature, wind velocity and low-humidity increased the prevalence of DED, especially for women and old people [11]. Despite less than 10% of time spent outdoors for these people, air pollutants exposure may be undeniably hazardous for human ocular surface and conjunctivitis [11][12][13]. A mouse dry eye model showed PM 10 induced apoptosis by increasing TNF-α, NF-κB-p65 and NF-κB expressions in the cornea, and abnormal differentiation and proliferation of the ocular surface [14]. Thus, it is urgent to assess the association of exposure to a wide variety of individual-level air pollutants with DED of young in their respective xed area. Thus, we need to assess the association of exposure and the key component of air pollution in the development of DED, so as to develop more effective measures of prevention and intervention.
As social environmental factor, constitution is expressed by "Qi" in the theory of Traditional Chinese Medicine (TCM). Intriguingly, a multicenter cross-sectional study in Chinese young people, suggested that the occurrence of DED may be associated with Qi stagnation, which is known as one of eight imbalanced constitutions of healthy people in TCM [15,16]. Based on TCM theory, "Balance" is referred to as Qi (yin and yang) homeostasis, which ensures the mental and physical health in Chinese Han population [17]. "Qi" is understood as an invisible effect, which corresponds to energy or signaling in modern western medicine. As for "Qi stagnation", it is regarded as a body state in which energy metabolism or signal transduction is blocked [18]. As an increased psychosocial stress, the prevalence of "Qi stagnation" is likely to increase among Chinese young people [15,19], yet the condition is often underrecognized and undertreated [16]. However, few studies have examined the possible interaction between air pollution and "Qi stagnation" on DED. The aim of this study was to explore whether air pollutants including particulate matter (PM 2.5 and PM 10 ), and gaseous pollutants (O 3 , SO 2 , CO and NO 2 ) is associated with increased risk of DED in a well-de ned population of Qi stagnation.

Study design and participants
We conducted a nested case-control study of the association between air pollution and dry eye disease (DED) at 17-20 years of age from He University students. He University is the third largest private medical college in China with an undergraduate population of 8,412 (regular) from 23 provinces in China, which is located in the beautiful Qipanshan forest scenic area of Shenyang, with the best air quality all year round (PM 2.5 : 4μg/m 3 , PM 10  coronaviruses, all students were forced to go home on holiday from January to July in 2020, and could not leave their respective communities under the control of the government. Until August 2020, most students were allowed to return to university. Thus, these students were exposed to air pollution from January to July in 2020, in their respective xed areas. We included students who lived in Liaoning and had no DED from the HEH screening visit in September 2019. We excluded students who had positive history of smoking, refractive surgery or contact lens wear, who used systemic medication, such as antianxiety, antidepressive, and antihypertensive medications. The enrolled students (n = 6407) read and signed an information and consent form before participating in this study, and attended the physical examination and completed the symptom questionnaires of the demographic information and daily custom, Qi stagnation and Ocular Surface Disease Index (OSDI). Then, we recruited 576 participants with DED (cases) and 576 without DED (controls) from 6407 students. Cases had to have at least two of the the following three signs in at least one eye: a) corneal uorescein staining (CFS), b) tear break-up time (TBUT), c) Schirmer's test with anesthesia. Cases included all patients newly diagnosed with active DED. Controls were selected from enrolled students without DED, were matched to cases (1:1) by age and sex. Individual air pollution exposure of participants (n = 1152) was estimated using the measurements from the municipal air monitoring station, which was located 1000 meters from each participant's home.

CFS
CFS was measured by the National Eye Institute/Industry (NEI) grading scale as previously described [20]. The total CFS score ranges from 0 to 15, with higher scores indicating greater abnormality. The NEI grading scale consists of a grid that divides the corneal area into ve sections, each of which is assigned a score between 0 and 3 depending of the amount and distribution of CFS.

TBUT
TBUT was performed using Lowther technique in each eye following a previous method [21]. A sterile uorescein strip (Indicator, He eye Co., Shenyang, China) was moistened using nonpreserved saline. The time between a blink and the appearance of gaps in the tear lm was tested in seconds. The time ranges from 0 to 7 seconds, with shorter times indicating greater abnormality.

Schirmer's test
After TBUT, the participant waited for 30 min, then performed Schirmer's test with anesthetic as previously described [22]. The Schirmer's test is measuring the total tear secretion (basic and re ex) in each eye, ranging from 1 to 7 mm in 5 minutes . The length of wetting of paper strips placed in the inferior cul de sac of the lower eyelid, with shorter lengths indicating greater abnormality.

Exposure assessment
The average ambient concentrations of PM 2.5 , PM 10 , O 3 , SO 2 , CO and NO 2 were obtained as previously described [23]. We obtained the average daily concentrations of these pollutants for each participant during seven months (from January 1, 2020 to July 31, 2020). The data were based on one of air pollution monitoring stations located within 1000 meters residence from each participant's home, and obtained using air pollution monitoring stations from National Climate Center in Ministry of Ecology and Environment of China by the use of AirData (http://www.zhb.gov.cn/). We estimated the average daily exposure for usual residence of each study participant, and then added the data together to calculate the monthly mean concentrations of the aggregate exposures. In addition, we calculated average 8-hr concentrations of CO and O 3 from these same monitors closest to each participant's home address.
Because the participants were in own closed communities, we ensured that these average values were available during this study period.

Confounders
We assessed potential individual-level confounders and effect measure modi ers, including demographics and socioeconomic characteristics (eg, age, sex, body-mass index, parents' education level), lifestyle factors (eg, myopia, time spent squinting at computer and cell phone, sleep, humidity), and "constitution" (de ned as the unique TCM entity exhibiting individual's biological feature and clinical presentation). To explore the potential role of "Qi stagnation" in the association between air pollution and DED, this variable was constructed from both behavioural and temperamental information estimated.

Statistical analysis
We used Chi-square test to compare the frequencies of baseline characteristics to the status of participants with DED and controls, usedtwo-sample t-test to compare the values of OSDI score, corneal staining score, TBUT and the result on Schirmer's test between two groups, and used 1:1 paired logistic models adjusting for all matching factors (age, and sex), to assess other in uencing factors for the association between DED and each average air pollutant concentration (PM 2.5 , PM 10 , O 3 , SO 2 , CO and NO 2 ) assigned to each participant for the 7 months before diagnosis (study entry) date. For ease of interpretation, the "Time" was recorded (1 = cases, 2 = age and gender-matched controls), and the "Status" was recorded [1 = case group (virtual censoring group), 0 = control group]. The paired pair number ("ID") was selected into "Strata" list box, and the in uencing factors were selected into "Covariates" list box. We selected exposure level below the 20th percentile as the referent category (baseline level) for each pollutant.
Potential confounders were based either on actual articles from the medical literature [11,[24][25][26]. We assessed potential confounders using a > 10% change for these risk factors such as, body-mass index (BMI), time spent at computer or mobile phone, myopia, parental education level, sleep, humidity, which could potentially confound the relationship between air pollution and DED. There were no factors confounding the associations between air pollution and DED. Thus, the nal analysis did not include the above factors. The matching factors were age and sex in this nal model. Furthermore, we assessed potential effect modi er (constitution) using likelihood ratio tests (LRTs) [27]. If the LRT differed at a signi cance level of < 0.10, we considered effect modi cation to be present. The effect modi ers according with this criterion included constitution. Data were analyzed with the use of SPSS software for Windows, version 20.0 (IBM SPSS, Inc., Chicago, IL). Table 2 showed ocular test values at baseline and screening. Table 3 summarized the basic characteristics of DED and matched controls nested within the college students. We identi ed 576 cases of DED, and 576 controls matched 1:1 to cases by age and sex, which were selected from the enrolled (September 2019) in He University during the study period. Less than one-fth (186) of the participants were classi ed as Qi stagnation. The study population was split by sex ratio according to 1:3 (male:female) between cases and controls. Although there was no signi cant difference in the ratio of Qi stagnation between males and females from controls (23.1% vs. 18.7%), a larger percentage of males were considered Qi stagnation than females among cases (21.1% vs. 9.5%). Table 4 showed that average ambient air pollution concentrations in the 7 months before the DED diagnosis date varied for participants. There were no signi cant differences in the Median air pollution concentrations between the Stagnant qi constitution and Balanced constitution for all pollutants. Table 5 present Spearman's correlation air-pollutant averages in the 7 months. The correlation coe cients showed only moderate correlations observed between air pollutants. The strongest correlation among coe cients was seen for PM 2.5 and PM 10 (r = 0.825). Table 6 demonstrated that the single and multi-pollutant models associations between air pollution (in quintiles) and DED diagnosis in the 7 months prior to diagnosis date/study entry among all cases and matched controls by the odds ratios (ORs) and 95% con dence intervals (CIs). The matching factors of age and sex were used to adjust all effect estimates.

DED
In single-pollutant models, DED was not associated with the criteria pollutant SO 2 . Additionally, PM 2.5 , Simultaneously, in multi-pollutant models, these results were generally consistent with those of the singlepollutant models, but most of effect estimates were attenuated. In contrast, for PM 2.5 and CO, effect estimates were enhanced. Especially, in multi-pollutant models, the OR for the 20th to 40th percent of quintiles of PM 2.5 exposure was 4.79 (95% CI: 1.94, 11.82).

Discussion
We provided a nested case-control study of its kind to assess the potential associations between air pollutants and DED within Chinese young people aged 17-20 years. The main strength of our study was that physical-social environment interactions between Stagnant qi constitution and the key component of air pollutants (PM 10 and CO) on the risk of DED. The risks of other air pollutants on DED were mutually adjusted to estimate their independent associations by using multi-pollutant model. This multi-pollutant model was used to explain the mutual confounding between pollutants, but, if these multi-pollutants were highly correlated, a number of potential variance in ation and bias is often possible [27,28]. We provided both single-and multi-pollutant model results, due to uncertainty about which model inducing the leastbiased estimates. In this sample, a dose-response relationships were observed between air pollutants exposure [CO and NO 2 (positive), O 3 (inverse)] and DED odds. But, no consistent association was observed between PM 2.5 , PM 10 , SO 2 and DED., which could be related to perturbing mechanisms of tear homeostasis [8].
Although we observed no consistent association between PM and DED, we found a positive correlation between PM exposure and DED. Moreover the correlation between PM 2.5 and DED was signi cantly compared with PM 10 exposure. This could be the reason that polycyclic aromatic hydrocarbon (PAH) level in PM 2.5 from indoor dust at Chinese public places accounted for 71.5% of total particulate PAHs [29].
Ambient particle concentrations are primarily from combustion products (coal combustion and biomass burning), secondarily from diesel vehicle emissions [30]. Recently, an experimental study has shown that PM 2.5 exposure slightly inhibit autophagy in the early stage, but activated autophagy in the late stage by homeostatic alteration between accelerating apoptosis and inhibit proliferation in human corneal epithelial cells [10]. Experimental studies have also shown that PM signi cant impaired human ocular health. It is the reason that heavy metals and PAHs are not only adsorbed on the surface of PM but also deeply encrusted in its structure [31,32]. The results of a previous study of combustion products were consistent with the hypothesis that diesel exhaust particles may be linked to cytotoxicity and in ammatory responses because both have induced oxidative stress in human conjunctiva cells [33]. Also, we noticed that our analysis for PM 2.5 and PM 10 was not consistent with one previous human epidemiological study reporting no association between short-term PM exposure and DED in the multivariate conditional logistic regression analyses [34], and suggesting the complex sources may represent PM 2.5 and PM 10 as nonspeci c pollutants, and leading to a failure in the relevance.
The positive associations between exposure to CO (or NO 2 ) and DED were consistent with other available evidence. Ambient CO and NO together in China in this setting are considered a surrogate of tra c emission, due largely to incomplete combustion inside motor engines [34]. NO 2 itself could acidify the tears and caused a decrease in tear break-up time [35,36], moreover, CO inhalation could lead to an increase in arterial and venous diameters, retinal blood ow velocity, and fundus pulsation amplitude in a placebo-controlled study [37].
Also, we observed a consistent inverse association between O 3 and DED, which is not consistent with the previous study, which reported increased O 3 levels were associated with dry eye symptoms in the Korean population [5]. It is the reason that ambient O 3 and NO 2 levels are bonded by chemical coupling, and the formation of O 3 needs to deplete nitrogen with the help of sunlight [38,39]. Consequently, when exposed to high levels of NO 2 , the individuals usually undergoes a low exposure to O 3 , and especially along hightra c areas [40].
There was no consistent association between SO 2 and DED in this study, which was consistent with other epidemiological studies [5,34]. These studies also estimated that SO 2 was not associated with symptoms or diagnosis of DED. This was related to the decreasing SO 2 levels, due to the worldwide use of fuel with low sulfur content [41].
To reduce differential misclassi cation of "Qi stagnation" to some extent, we blindly assigned "Qi stagnation" status to case and air pollution exposure status. In addition, for study eligibility, we matched cases and controls on age and sex, and "Qi stagnation" status were diagnosed before recorded air pollution exposure. Moreover, a person's Qi state is very di cult to change in a short time [16].
Limitations to the present study include the small sample size. This raised the possibility that ndings were attributed to chance and hence need to be replicated. We did not consider seasonal and daily variations in air pollutants, and not collect other potential confounders, such as wind and temperature.
Because of the severe air pollution in China, our ndings were probably not well-suited for re ecting the integrated population from developed countries in North America and parts of Europe, due to their good air quality. Nevertheless, our analysis may provide valuable references for some developing countries. Heavy air pollution caused DED, which may lead symptom like Qi stagnation indirectly. Thus, we need to indicate causal relationships between Qi stagnation and exposure of air pollutants and DED.
In summary, our results showed positive associations between PM 2.5 , PM 10 , CO, NO 2 and risk of DED. The Stagnant qi constitution might increase the association between PM 10 , CO and DED.

Conclusion
Our prospective cohort study suggested that Stagnant qi constitution might increase the susceptibility short-term exposure (PM 10  Informed consent was obtained from all parent and/or legal guardian of minors (under 18).

Consent for publication
Not applicable.

Availability of supporting data
The datasets generated during and analysed during the current study are available from the corresponding author on reasonable request for all interested researchers upon requests sent to the author's o ce. The initial contact for request should be addressed to the corresponding author's institution.

Competing interests
The authors declare that they have no con icts of interest.

Funding
This study was funded by Liaoning natural science foundation (grant number 20180550900).