Association between Air Pollution Exposure and Daily Outpatient Visits for Dry Eye Disease: A Time-Series Study in Urumqi, China

Round 1

Reviewer 1 Report

I have had the opportunity to review the manuscript entitled “Association between air pollution exposure and daily outpatient visits for dry eye disease: a time-series study in Urumqi, China”. The subject of this study is interesting and limited evidence is available about this association. The authors attempted to report the study clearly in all sections. I have a few comments that may be improved the reporting of the study. 

1. Introduction:

 I think that the first paragraph of the introduction is not needed to be mentioned. It is duplicate information since the authors describe the study location conditions in the method section.      

1. Methods

1.2. The retrospective ecological time-series design has the inherent limitation to investigate a cause-effect relationship. In this design, the estimating of air pollutants and metrological paramours is not based on individual exposure. Maybe, a perspective time-to-event study design is a better alternative approach. Please consider this limitation in the discussion.

2.2. The diagnosis of dry eye disease was based on the electronic information system. These data may be associated with misclassification bias due to errors when entering the data into the system. Is there any information about the accuracy of these data? Please clarify or mention this in the limitation section. 

3.2. The confounding effects due to lifestyle factors such as sleep status and electronic device use and some occupational exposures such as night shift workers and visual display terminal workers can influence the estimated effect size. Please consider this in the limitation section.   

4.2. Unfortunately, I don’t see the supplemental files that the authors mentioned in the manuscript. Please clarify. 

5.2. Some statistical models have been employed in previous studies that can be used for estimating the association by considering all air pollutants and metrological parameters. Please see this publication (Zhang Y, Dong T, Hu W, Wang X, Xu B, Lin Z, Hofer T, Stefanoff P, Chen Y, Wang X, Xia Y. Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: Comparison of three statistical models. Environ Int. 2019 Feb; 123:325-336. doi: 10.1016/j.envint.2018.11.076. Epub 2018 Dec 14. PMID: 30557812.)

Author Response

Response to Reviewer Comments

Dear Reviewer 1.
Thank you very much for your meticulous comments and suggestions. These comments are valuable and helpful for us to revise and improve the paper, and they are also important guidance for our research. We have carefully studied these comments and made revisions, which we hope will be approved by you. The point-by-point revised parts are marked in red as follows.

1.Introduction:

I think that the first paragraph of the introduction is not needed to be mentioned. It is duplicate information since the authors describe the study location conditions in the method section.      

Response: Thank you very much for your comments. Based on your suggestions, we have removed the repetitive content of the first paragraph and added the necessary critical information to the last paragraph of the introduction section to make the logic complete and smooth. Thank you again for your enthusiastic work!

1.Methods

1.2. The retrospective ecological time-series design has the inherent limitation to investigate a cause-effect relationship. In this design, the estimating of air pollutants and metrological paramours is not based on individual exposure. Maybe, a perspective time-to-event study design is a better alternative approach. Please consider this limitation in the discussion.

Response: We appreciate your comments. All our authors agree that your proposed perspective time-to-event study design is indeed a better approach to investigate the relationship between air pollution and meteorological factors and health, and we have described the limitations in the "Discussion" section as you suggested, as follows: "Second, although we used both the current and permanent postal addresses in the medical records to determine the exposure addresses, and multiple procedures were also used to screen as many patients as possible who lived in the major areas of Urumqi, the data from fixed monitoring stations were closest to the individual’s residence. However, due to misregistration of information, there may still be postal address zip codes that differ from the actual area of residence and other occurrences that may undermine the validity of the study, and individual-level exposure and indoor contamination cannot be examined in detail. In addition, given the inherent limitations of the perspective retrospective ecological time-series design in investigating causality, especially since estimates of air pollutants and other measured parameters were not based on precise individual exposures, a time-to-event (TTE) study design may be a better alternative, because it not only took into account the occurrence of the event but can also include the time of the event, future designs based on mobile monitoring systems will be expected to provide accurate individual-level exposures as well as indoor pollution, which were currently unavailable for detailed examination." Thank you again for your valuable comments.

2.2. The diagnosis of dry eye disease was based on the electronic information system. These data may be associated with misclassification bias due to errors when entering the data into the system. Is there any information about the accuracy of these data? Please clarify or mention this in the limitation section. 

Response: Thanks very much for your comment. As your suggestion, we have added a description of the accuracy of the diagnosis of DED and the data source of the relevant electronic information system in the “Materials and Methods” section, as follows: "In addition, we examined the diagnostic entries for DED in the outpatient medical records and the corresponding chief complaints and ophthalmologic examinations. The final diagnostic information included was reviewed by physicians with intermediate or higher qualifications in the department or by the corresponding attending physicians to ensure the accuracy of the diagnosis and data classification. The geographical location of the hospital is shown in Figure 1 and Supplementary Material B.". The limitations of the study are also described in the section of "Discussion", described as follows: "Third, the diagnosis of DED was relied on electronic medical record information systems, and despite appropriate review procedures were adopted, these data may be associated with misclassification bias due to errors when entering the data into the system. On the other hand, electronic medical record systems lacked some valuable information, like socioeconomic status, education level or occupation, and DED subtypes, thus preventing us from further analyzing those effect of individual differences.". Thank you for your enthusiastic comments on improving our work.

3.2. The confounding effects due to lifestyle factors such as sleep status and electronic device use and some occupational exposures such as night shift workers and visual display terminal workers can influence the estimated effect size. Please consider this in the limitation section.   

Response: Thanks for your helpful comment. Given your suggestion, we have added confounding factors due to lifestyle factors and some occupational exposures that can affect the estimated effect size to the limitations section of the “Discussion”, which are described as follows: "Fourth, multiple lifestyle factors, such as sleep status and electronic device use, as well as some occupational exposures, such as night shift workers and visual display terminal workers, may affect the estimated effect sizes. In addition, UV exposure, allergens such as pollen, atmospheric suspended particles or solvents (e.g. volatile chemicals) may also act as potential confounders in this study. As a result, future studies should take these difficult-to-measure confounding factors into full account". Thanks again for your advice!

4.2. Unfortunately, I don’t see the supplemental files that the authors mentioned in the manuscript. Please clarify. 

Response: We greatly appreciate your comment and apologize for our oversight. We have revisited the submission and uploaded all supplementary files to ensure that all information submitted is complete. Thank you again for your kind response.

5.2. Some statistical models have been employed in previous studies that can be used for estimating the association by considering all air pollutants and metrological parameters. Please see this publication (Zhang Y, Dong T, Hu W, Wang X, Xu B, Lin Z, Hofer T, Stefanoff P, Chen Y, Wang X, Xia Y. Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: Comparison of three statistical models. Environ Int. 2019 Feb; 123:325-336. doi: 10.1016/j.envint.2018.11.076. Epub 2018 Dec 14. PMID: 30557812.)

Response: We appreciate your valuable comments. We thoroughly reviewed and studied the articles you recommended, which gave us with more insight.

In this study, they considered the results of three different statistical methods (i.e., generalized linear regression models, WQS regression models, and BKMR models) to examine the effects of seven relevant environmental chemicals on obesity in the general U.S. population. In particular, this study emphasizes the assessment of the joint effects of chemicals on health outcomes through the use of different statistical methods, provides a descriptive summary and comparison of the results of these three statistical models, fully discusses the advantages and weaknesses of specific methods, and is a useful reference for methodological choices in environmentally relevant health research.

Generalized linear regression models, including multivariate logistic regression and linear regression, are among the most commonly used statistical models for assessing the health effects of chemicals and are the main methodological models we refer to in this study [1,2,3,4]. However, this model has some limitations. Firstly, in some cases we need to consider mixed environmental exposures and their complex non-linear interactions to investigate specific causal relationships [2,5], but this model may ignore the combined effects of other chemicals, which may lead to false-positive or false-negative results [6]. On the other hand, it is incorrect to include all chemicals of interest in one generalized linear regression model, as the high correlation between chemicals may lead to distorted results [7]. Moreover, it is impossible to analyze their specific chemical interactions in a simple model.

The WQS and BKMR are two recently developed models for analyzing the health effects of chemical mixtures. Their advantage is the ability to allow simultaneous consideration of a range of chemicals with high correlations. The WQS regression model examines the whole-body burden of chemical exposures based on empirically determined weights derived from self-sampling. This approach better simulates the complexities of real-life exposures and, as a result, WQS regression models are more sensitive than single chemical analyses in identifying important factors [8,9]. However, similarly, an important limitation of the WQS is that the joint effects of chemicals with different impact directions cannot be assessed simultaneously.

BKMR analysis can identify nonlinear effects and chemical interactions [10]. In contrast to WQS regression models, BKMR analysis can capture exposure-response relationships with other chemicals fixed at certain levels. In addition, analyses using BKMR can assess the systemic effects of chemicals with different effect directions and can also detect interactions between each of two kinds of these chemicals. However, one limitation of the BKMR model comes from its kernel algorithm. By fixing other chemicals at a certain level to infer the exposure-response function, this prevents estimation of the effects of a common exposure pattern of high and low levels of chemicals.

Therefore, the combined use of both methods, WQS and BKMR models, allows to fully take into account their advantages and disadvantages in order to better investigate the detailed interactions between chemical mixtures.

In this study, we used a quasi-Poisson generalized linear regression model with a distributed lagged nonlinear model (DLNM) to fit the effect of exposure to air pollutants on DED visits, considering that daily visits to DEDs are considered rare events that approximate a quasi-Poisson distribution. DLNM is often used for data analysis with time-lagged effects, such as in epidemiological studies, where the lagged effects of changes in factors such as temperature often need to be taken into account and DLNM is introduced for modeling analysis. Lagged effects refer to the correlation between the values of the dependent variable and the past values of the independent variable. DLNM represents a modeling framework that can flexibly describe associations that show potential nonlinearities and lagged effects in time-series data. The methodology is based on the definition of cross bases, which are two-dimensional function spaces represented by the combination of two sets of based functions that specify the relationship between predictor and lagged variables, respectively. Given a defined temporal structure of the data and a simple definition of the lag dimension, time-series study designs can offer a variety of advantages to deal with lag effects, where time divisions are directly specified by equally spaced and ordered time points. In this case, lagged effects can be elegantly described by the DLNM, whereby multiple parameters can be used to account for effects at different time lags, thus distributing the effects of individual exposure events over specific time periods. Such statistical models as well as processing methods we used have been widely applied to quantify health effects in recent studies of environmental factors such as air pollution, meteorological factors, etc. [11,12], and is commonly recognised as reliable and valid.

Overall, generalized linear regression can provide a simple relationship between a single chemical and an outcome in environmental epidemiology studies, thus distributing the effects of a single exposure event over a specific time period. The WQS model can investigate the effects of mixed exposure burdens on outcomes in only one direction at one time. The BKMR model can evaluate the exposure-response function of each chemical at some level while controlling for the others, as well as the interaction of any two chemicals. Considering that daily outpatient visits to the DED are considered rare events that approximate a quasi-Poisson distribution, the quasi-Poisson generalized linear regression model with distributed lagged nonlinear model (DLNM) that we employed can use multiple parameters to account for effects at different lags, thus distributing the effects of individual exposure events over specific time periods and flexibly describing associations exhibiting potential nonlinear and lagged effects in time-series data, which is widely recognized and utilized in environment-related health studies. As a result, these models evaluate different aspects, and future research should consider the joint interpretation to fully reveal their strengths and limitations in order to achieve functional complementarity.

In the "Discussion" section, we have collated the information above, added corresponding descriptions to the limitations paragraphs, and quoted the literature you suggested. For more information, please refer to the respective paragraphs in the revised paper, which we have marked. Thank you again for your recognition of our work and valuable comments on the enhancement of our manuscript!

Reference:

[1] Bhandari R, Xiao J, Shankar A. 2013. Urinary bisphenol a and obesity in U.S. Children. Am J Epidemiol 177:1263-1270

[2] Kim S, Kim S, Won S, Choi K. 2017. Considering common sources of exposure in association studies - urinary benzophenone-3 and dehp metabolites are associated with altered thyroid hormone balance in the nhanes 2007-2008. Environ Int 107:25-32

[3] Wei Y, Zhu J, Nguyen A. 2014. Urinary concentrations of dichlorophenol pesticides and obesity among adult participants in the u.S. National health and nutrition examination survey (nhanes) 2005-2008. Int J Hyg Environ Health 217:294-299

[4] Ye X, Wong LY, Zhou X, Calafat AM. 2014. Urinary concentrations of 2,4-dichlorophenol and 2,5-dichlorophenol in the U.S. Population (national health and nutrition examination survey, 2003-2010): Trends and predictors. Environ Health Perspect 122:351-355

[5] Valeri L, Mazumdar MM, Bobb JF, Claus Henn B, Rodrigues E, Sharif OIA, et al. 2017. The joint effect of prenatal exposure to metal mixtures on neurodevelopmental outcomes at 20-40 months of age: Evidence from rural bangladesh. Environ Health Perspect 125:067015

[6] Czarnota J, Gennings C, Colt JS, De Roos AJ, Cerhan JR, Severson RK, et al. 2015. Analysis of environmental chemical mixtures and non-hodgkin lymphoma risk in the nci-seer nhl study. Environ Health Perspect 123:965-970

[7] Marill KA. 2004. Advanced statistics: Linear regression, part i: Simple linear regression. Acad Emerg Med 11:87-93.

[8] Artacho-Cordon F, Leon J, Saenz JM, Fernandez MF, Martin-Olmedo P, Olea N, et al. 2016. Contribution of persistent organic pollutant exposure to the adipose tissue oxidative microenvironment in an adult cohort: A multipollutant approach. Environ Sci Technol 50:13529-13538

[9] Nieves JW, Gennings C, Factor-Litvak P, Hupf J, Singleton J, Sharf V, et al. 2016. Association between dietary intake and function in amyotrophic lateral sclerosis. JAMA Neurol 73:1425-1432

[10] Bobb JF, Valeri L, Claus Henn B, Christiani DC, Wright RO, Mazumdar M, et al. 2015. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics 16:493-508

[11] Muscogiuri G, Barrea L, Laudisio D, Savastano S, Colao A. 2017. Obesogenic endocrine disruptors and obesity: Myths and truths. Arch Toxicol 91:3469-3475.

[12] Scinicariello F, Buser MC. 2016. Serum testosterone concentrations and urinary bisphenol a, benzophenone-3, triclosan, and paraben levels in male and female children and adolescents: Nhanes 2011-2012. Environ Health Perspect 124:1898-1904

Best wishes,

Zheng-Xuan Jiang

Author Response File: Author Response.docx

Reviewer 2 Report

The manuscript addresses an important gap. There is a need however to address several limitations in the representation and organization of the manuscript as per the attached comments.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments

Dear Reviewer 2.
Thank you very much for your enthusiastic work and meticulous comments. These comments are valuable and helpful for us to revise and improve the paper, and they are also important guidance for our research. We have carefully studied these comments and made revisions, which we hope will be approved by you. The point-by-point revised parts are marked in red as follows:

General:

Use of long sentences, repetition and placement of various texts in the wrong section undermines the flow and greatly undermines the logical flow, clarity and conciseness of the manuscript. The authors should put considerable effort in this section. Other comments are as follows

1.   General observations

•   Abstracts include the software used in the analysis

Response: Thank you very much for your comment. We have added information about the software we used to the abstract section as your comment suggests.

•   The Statement“The harsh climatic conditions have resulted in Urumqi covering a vast area but with few habitable sites, and”on line 42 could be deleted

Response: Thank you very much for your comment. We have removed the relevant statement as you suggested. Thank you again for your comment.

•   Line 67could start with Some systematic and meta-analysis

Response: Thank you very much for your suggestions. We have modified the structure of the corresponding paragraph as follows: "Some systematic review and meta-analysis reported that the prevalence in western and northern China was significantly higher than in the southeastern area, which strongly suggests the potential importance of regional factors. DED has become a growing public health problem worldwide, particularly in China, due to its impact on daily activities and quality of life [1,2] ". Thank you very much for your comment.

•   Use of long sentences 53-57; 115- 116; 82-85; 107- 113; 120- 131 and several other sections undermines the quality of the manuscript

Response: Thank you very much for your comment. We have carefully considered your comments and have reviewed all the sentences in the article, as shown in the article's corresponding paragraphs, we replaced long sentences with shorter ones to make the content more fluid and easier to understand, which we have marked in the revised manuscript. Thank you again for your valuable comments to improve the quality of our manuscript!

•   repetitiveness of some words eg. Region on 68-69; 107- 113; 120- 131

Response: Thank you sincerely for your comment. All of our authors agreed with your comments and carefully revised the words and sentences that were used repeatedly, as shown in the corresponding paragraphs, which were marked in the revised manuscript. Thank you for your valuable comments on improving the quality of our manuscript. 

•   industrial structure of the region is heavily industrial in line 103- 104 could be deleted as it adds no value/ makes the text

Response: Thank you very much for your comment. We have removed the corresponding sentences based on your comments. Please check the relevant paragraphs in the article, which have been marked in the revised version. Thank you very much for your comment.

•   Line 70 “where some”could be deleted. Review Line 68-72 ; Line 77-80 to avoid unnecessary repetition; 82-85; 86-92 (use short sentences)

Response: Thank you very much for your comment. In accordance with your suggestions, we have checked and corrected the corresponding sentences and paragraphs individually. Please check the relevant marked content in our revised article. Thank you again for your valuable suggestions to improve our manuscripts.

•   Line 136- 142 decreases the clarity of the figure and should appear as key/ legend to the map/ figure and not as text

Response: Thank you very much for your comment. We have adjusted Figure 1 and the related legend content according to your comments, please check the corresponding section within the revised article. Once again, thank you very much for your comments and enthusiastic suggestions for improving the quality of our manuscripts.

•   383-394 could be made more concise

Response: Thank you very much for your comment. We have revised the manuscript based on your comments and made the presentation more concise. Thank you for your valuable comments.

2.   Specific comments

2.1.Organisation of the manuscript

•   Several sections are in the wrong place. For example, a lot of materials in the discussion section should be in the background/Introductory section. Check for example line; 290-352. After correcting for this, delete repetitions

Response: Thank you very much for your valuable comments. Based on your comments, we have carefully reviewed the relevant paragraphs in the Introduction and Discussion sections, checked for relevant repetitive descriptions, and deleted or restructured, please check the revised manuscript. Thank you again for your enthusiastic work and for your comments on improving our manuscript.

•   Line 148- 152; 159- 164, revise the statements and include relevant references from previous studies as you have stated

Response: We greatly appreciate your comments. Given your comments, we have added relevant references in these two sections. Please review the sections that we have revised and marked. Thank you again for your valuable comments.

•   Number the equation in line 170

Response: Thank you very much for your valuable comments. We have numbered the formulas you have mentioned, please check the revised version. Thank you again for your valuable comments.

•   Line 180- 181 is hanging. It could be moved to relevant section.

Response: Thank you very much for your valuable comments. We checked the relevant content and revised the paragraphs accordingly. Please check the revised manuscript and thank you again for your comments.

•   Line 395-406 could be misplaced. Its better represented in results section

Response: Thank you for your comment. We have carefully reviewed the corresponding paragraphs and have moved the content to the last paragraph of the results section based on your comments. Please review the corresponding changes and thank you again for your enthusiastic work and detailed guidance.

•   Check on referencing style in line 408- 420 and revise to present concise statements

Response: Thank you for your valuable comments. According to your comments, we have re-examined the corresponding paragraphs you mentioned and revised the style of citing references to make the presentation of the content more concise and clearer. Thank you again for your valuable comments on the organization and presentation of our manuscript.

•   Line 453-462 should be the basis for you justifying why you choose the Ophthalmology Department of the First Affiliated Hospital of Xinjiang Medical University and not any other hospital data set. This should come out clearly in the relevant section.

Response: Thank you for your comment. We carefully considered your comments and unanimously agreed with your viewpoint, and we reorganized the relevant statements and added more details to the "Materials and Methods" section on why we chose the Ophthalmology Department of the First Affiliated Hospital of Xinjiang Medical University to make the paragraph more clearly expressed. The details are as follows: "The hospital has the largest and most modern ophthalmology clinic in Xinjiang province. As a hospital under the supervision of the Chinese Ministry of Health, it is considered "one of the top ten in China and the first in Northwest China" and has received many awards and general recognition from medical colleagues. It is the leader in the treatment of ocular surface diseases such as DED in Xinjiang province, and it has conducted many epidemiological surveys of eye diseases in the province. As a result, it has become the preferred and most favored medical institution for patients with eye diseases in Xinjiang and surrounding areas, including Urumqi.". Thank you again for your valuable comments on our manuscript.

•   Is validation necessary in similar studies to what you have analysed.? If yes, then is lack of validation (Line 462-468) one of the study limitations especially in absence of assumptions underpinning your study?

Response: Thank you for your comment, regarding your mention of validation in the same type of study, all our author members discussed and finally came to a conclusion. Considering that we conducted the first study on the association between air pollution and DED in the Urumqi region, it was not possible to find a valid validation of the results of similar related studies conducted in this same region. The same type of studies have been published in other regions, but considering that the number is still limited and that the effects of air pollution on DED are of increasing concern, it is still necessary to conduct similar validation of the limited findings to obtain the most up-to-date and comprehensive epidemiological evidence on the effects of different types of air pollutants on DED, which is considered meaningful and essential to some extent. As we discussed in the original manuscript Line 462-468. However, as you mentioned, the lack of assumptions based on the region underpinning our study remains a non-negligible limitation of our study. We have added descriptions to the limitations paragraphs in the “Discussion” section in accordance with your comments, please check the corresponding sections marked in the revised manuscript for details. Once again, thank you very much for your valuable comments on improving the quality of our manuscripts.

2.2. Methods and Materials

Some of the flaws/ Missing explanations in this section are as follows.

•   What are the assumptions for the study? This need to be well explained otherwise the reader may make wrong judgement on the reliability and validity of the study

Response: Thank you for your comments. We have explained the hypotheses of this study in detail according to your comments and organized them in the relevant paragraphs of the “Materials and Methods” section, as follows: “Our research hypothesis is based on a few similar studies of the same type that have been published, in which the type and level of air pollution were associated with the onset of DED. Increased levels of six major air pollutants (PM₁₀, PM_2.5, SO₂, NO₂, O₃, and CO) may increase the risk of DED outpatient visits, and this correlation has lag effects, while meteorological factors were one of the important confounding factors.” and the description of the assumptions for the selection of the statistical model and the determination of the relevant parameters. Please review the corresponding section in the revised manuscript and thank you for your valuable comments.

•   What are the likely confounding factors and how did you control for them?

Response: Thank you very much for your comments. First, as mentioned in the "Materials and Methods" section, multiple meteorological factors are the main confounders in this study; thus, we control the effects of the main meteorological confounding variables, including relative humidity, mean temperature, and air pressure, in all models by smoothing the natural cubic spline curves (ns) using three degrees of freedom (df), as described in the model construction paragraphs.

Furthermore, UV irradiation, allergens such as pollen, and certain atmospheric suspended particles or solvents (e.g. volatile chemicals) that are difficult to measure may be confounding, and because these factors are not captured by conventional assays and thus difficult to control, we have described them as a limitation of this study in the “Discussion” section. Thank you again for your valuable comments.

•   Why didn’t you consider some form of sensitivity analysis?

Response: Thank you for your valuable comments. All of us authors carefully considered your comments and agreed that sensitivity analysis is feasible and necessary, and we conducted sensitivity analyses in the revised version of the manuscript for different years of data to verify the study's stability in accordance with your comments. Specifically, we conducted sensitivity analyses for 2013-2017, 2013-2018, and 2013-2019 with this study (for 2013-2020), and the results suggest that our findings remain stable across different years. For details, please see the new marked corresponding section in the results paragraph in the text and Supplementary Material B. Thank you again for your comments on improving the quality of our manuscript.

•   Could the implied use of zip code as confirmation of residence (line 113) undermine validity of the study? (unless otherwise; someone could be having a postal address zip that is different from the actual residential area)

Response: Thank you for your valuable comments. First of all, as you mentioned, we must acknowledge that there may be people with postal address zip codes different from their actual area of residence and other occurrences that may undermine the validity of the study due to incorrect information registration. However, it must be noted that in this study, we collected both the current and the permanent residence zip codes in the medical record information, and we prioritized the inclusion of the current residence address for the follow-up study in case of discrepancies between the two information, to guarantee the validity of the study as much as possible. This method has been widely used in existing studies of the same type, such as [1,2,3]. To some extent, this can be considered reliable. We updated the "Materials and Methods" section to more clearly present the details of this data treatment, and we added the limitations paragraph to the "Discussion" section. Thank you again for your comments on improving the quality of our manuscript.

Reference:

[1] Bao N, Lu Y, Huang K, Gao X, Gui SY, Hu CY, et al. 2021. Association between short-term exposure to ambient nitrogen dioxide and the risk of conjunctivitis in Hefei, China: A time-series analysis. Environ Res 195:110807

[2] Cheng P, Liu C, Tu B, Zhang X, Chen F, Xu J, et al. 2022. Short-term effects of ambient ozone on the risk of conjunctivitis outpatient visits: A time-series analysis in Pudong new area, shanghai. Int J Environ Health Res:1-10

[3] Mu J, Zeng D, Fan J, Liu M, Yu S, Ding W, et al. 2021. Associations between air pollution exposure and daily pediatric outpatient visits for dry eye disease: A time-series study in Shenzhen, China. Int J Public Health 66:1604235.

2.3.Representation of the analysis

•   Some of the sections presented as texts e.g. Line 199-203; could better be presented as tables/ figures

Response: Thank you for your precious comment. We agree with your comments and have revised this section to "In general, during this eight-year period, there appears to be an annual cyclical trend for the six regional air pollutants and DED outpatient visits, with CO, SO₂, and O₃ falling below China's national ambient air quality standards, but NO₂, PM_2.5, and PM₁₀ exceeding these standards. Please see Table 1 for more information.". And we modified Table 1 to include this part of the description as a column in the new table, supplementing the Chinese national ambient air quality standard values column. Thank you again for your comments on improving the quality of our manuscript.

•   The use of Mean ±SD  notation for mean and standard deviation as a single column could improve the table /clarity of the results

Response: Thank you for your valued comment. As you suggested, we have included the mean and standard deviation as a column using the mean ± SD sign to improve the consistency of the results. Please check our updated table in the revision manuscript.

•   An introductory statement after line 216 (Refer the reader your to figures / tables before presenting the data/ writeup).This would improve the representation of your manuscript

Response: Thank you for your valuable comments. We have added the introductory statement "Please refer to the figure/table before presenting the data/writeup" after the endnote in Table 1 to improve the representation of our manuscript. Please review our revised Table 1. Thank you again for your valuable comments on the enhancement of our manuscript.

•   Line 232-234 not clear (contradicting) and needs to be corrected

Response: Thank you for your comment. We apologize for the lack of clarity and have revised the sentence in the manuscript to "CO and O₃ were not significantly associated with the risk of DED outpatient visits, while the remaining four air pollutants were revealed to have a weak positive association with the risk of DED.". Thank you again for your comments.

•   How logical is it for an increases on the stated variables on day 0 (line 234)

Response: Thank you for your precious comments. For the logic of the increases on the stated variables on day 0, our interpretation is:

On the one hand, from the modeling perspective, we defined the same-day exposure as lag 0 and the maximum lag days as 7 in order to assess the potential lagged effects of air pollutants. The day 0 effect value represents the correlation between same-day exposure and the number of matched DED outpatient visits.

On the other hand, from the pathogenic perspective, we analyzed the acute effects of air pollution by single-day lags (lag 0 to lag 7) and cumulative lags (lag 0-1 to lag 0-7), the effect of day 0 on the variables described can reflect the hyperacute effect of the day (within 24 hours) of the patient's exposure to air pollution (especially severe air pollution) that induces an outpatient visit for DED. This is widely used in the same type of research, such as [1,2,3].

Reference:

[1] Bao N, Lu Y, Huang K, Gao X, Gui SY, Hu CY, et al. 2021. Association between short-term exposure to ambient nitrogen dioxide and the risk of conjunctivitis in Hefei, China: A time-series analysis. Environ Res 195:110807

[2] Cheng P, Liu C, Tu B, Zhang X, Chen F, Xu J, et al. 2022. Short-term effects of ambient ozone on the risk of conjunctivitis outpatient visits: A time-series analysis in Pudong new area, shanghai. Int J Environ Health Res:1-10

[3] Fu Q, Mo Z, Lyu D, Zhang L, Qin Z, Tang Q, et al. 2017. Air pollution and outpatient visits for conjunctivitis: A case-crossover study in hangzhou, china. Environ Pollut 231:1344-1350

•   Scientifically, what do the confidence intervals given in line 238-242 imply (are they statistically significant or not)

Response: Thank you for your valuable comments. We apologize for the unclear presentation, and we have revised the statement in the corresponding paragraph you mentioned, and added the judgment about whether it is statistically significant or not. The revised new statement is as: "(RR values range 1.059-1.067)" was replaced with "(RR values range from a minimum of 1.059 to a maximum of 1.067, and all were statistically significant correlations)", "(RR values range 1.039-1.059)" was changed to "RR values range from a minimum of 1.039 to a maximum of 1.059 at lags 0-4 to 0-7 days, and all were statistically significant correlations", and "(RR=1.092, 1.005-1.186)" was changed to "(RR=1.092, 95% CI: 1.005-1.186), it was statistically significantly associated with DED at lags 0-2 days". Thank you again for your valuable comments to improve the quality of our manuscript.

•   Line 241-242 is hanging

Response: Thank you for your comments. We have made formatting-related adjustments to this section in accordance with your comments. Please check the revised manuscript. Thank you again for your comments.

•   Table 2 could better be represented as landscape

Response: Thank you for your comments. We have made relevant formatting adjustments to this section in accordance with your comments, and the revised table is marked in the text. Please check the revised manuscript. Thank you again for your comments.

•   Need to anchor your discussion/ findings in line 248-255 on existing knowledge / literature

Response: Thank you for your valuable comments. We have made the relevant formatting adjustments to this section in accordance with your comments, and the revised version is as follows: "Combining the results of the multi-pollutant model, we failed to find significant correlations between CO and O₃ and the risk of DED outpatient visits, but NO₂, SO₂, and PM_2.5 showed statistically significant correlations between per 10 μg/m³ increase and outpatient visits of DED patients. Furthermore, because the findings in single and multi-pollutant models were inconsistent, it was unclear whether there was a statistically significant correlation between increased PM₁₀ exposure and DED outpatient visits. Sensitivity analysis for different years tested these findings (Supplementary Materials B).". Thank you again for your enthusiastic work and for your comments on improving our manuscript.

2.4.Discussion

Need to be tied to existing body of knowledge, critical analysis and appropriate synthesis of the ideas presented (not necessarily word to word recasting from background section). After correction and removing background material to the relevant section, this section needs to improve to a great extent.

Response: Thank you for your enthusiastic work and valuable comments for improving the quality of our manuscript. In accordance with your comments, we have reviewed and reorganized the entire text, especially the discussion as well as the introduction paragraphs, correcting and removing repetitive background material, linking it as much as possible to the body of knowledge available and to the published findings, and critically analyzing and appropriately synthesizing the ideas presented in this study.

Once again, thank you sincerely for your enthusiastic work. We appreciate your detailed and valuable suggestions for the improvement of our manuscript, and we are always looking forward to your valuable comments！

Best wishes,

Zheng-Xuan Jiang

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

I am satisfied with the revised manuscript. However as suggested in the previous manuscript, check the use of "confirming" on line 322 as well as use the suggested ± notation for presentation of mean and SD in a single line.

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments (Round 2) 

Dear Reviewer 2.
Thanks very much for taking your time to review this manuscript. We really appreciate all your generous comments and suggestions! I apologize for our oversight in the previous revision, please find my second round of itemized responses in below and my revisions in the re-submitted files:

1. However as suggested in the previous manuscript, check the use of "confirming" on line 322

Response: We greatly appreciate your comment and apologize for our oversight. We have revised the word "confirming" to "We confirmed", please check our revised version in line 322. Thank you again for your comments！

2. as well as use the suggested ± notation for presentation of mean and SD in a single line.

Response: Thanks very much for your comment. We have checked the revised version of the manuscript and I apologize for our oversight in this comment. As you suggested, we  have included the mean and standard deviation as a column using the mean±SD sign to improve the consistency of the results. Please check our updated table in the revision manuscript.

Once again, we sincerely thank you for your impressive and enthusiastic work. I apologize for our oversight in the previous revision. We greatly appreciate your valuable and detailed suggestions for improving our manuscript! And we are always looking forward to your valuable comments！

Best wishes,

Zheng-Xuan Jiang

Author Response File: Author Response.docx