Prevalence of dry eye during the COVID-19 pandemic: A systematic review and meta-analysis

Objective During the COVID-19 pandemic, many people devoted longer time to screen viewing due to the need for study, work, and online social activities, instead of outdoor activities, which may have led to an increase in dry eye symptoms. This study aimed to evaluate the prevalence of dry eye during the COVID-19 pandemic. Methods PubMed, Cochrane Library, Embase, and Web of Science were searched from January 1, 2020 to October 20, 2022. Cross-sectional surveys on dry eye prevalence conducted after January 1, 2020 were included. Two review authors independently performed data extraction and assessed study quality. The random-effects model was used to analyze the prevalence of dry eye, and the odds ratio was used to assess the strength of the association between variables. Subgroup analysis was performed to detect heterogeneity, the leave-one-out method for sensitivity analysis, and the Egger test for publication bias. Results A total of eleven studies with 15692 individuals met the eligibility criteria. The prevalence of dry eye during the COVID-19 pandemic was 61.0% (95%CI: 51.8%-70.2%) globally and 56.7% (95%CI: 45.3%-68.1%) in Asia. The prevalence of dry eye had significant differences in sex and visual display time, with higher prevalence among females and visual display time of more than 4 hours per day. Subgroup analysis was performed based on diagnostic tools, study population, and average age. A significant difference was found in diagnostic tools, but no significant change in heterogeneity (P<0.05). The leave-one-out method showed stable results, and the Egger test identified no significant publication bias. Conclusion The prevalence of dry eye during the COVID-19 pandemic is significantly higher than before, and a higher prevalence is found among females and those having a visual display time of more than 4 hours per day.


Introduction
During the latter part of 2019, COVID-19 struck the world and significantly changed people's lifestyles: people reduced outdoor activities and spent more time on screens due to the lockdown measures [1].Recent epidemiological surveys have shown an increase in the prevalence of psychological disorders such as anxiety and depression during the COVID-19 pandemic [2,3], as well as a significant increase in the number of people suffering from digital eye strain due to the change of lifestyle [4][5][6].
It is well known that the prevalence of dry eye is associated with the use of video display units [7][8][9].Dry eye is characterized by a loss of the tear film homeostasis, accompanied by ocular symptoms including dryness, discomfort, pain, grittiness, blurry vision, foreign-body sensation, and visual disturbance [10].Not only does dry eye affect people's quality of life and reduce their work efficiency [11][12][13], but it also increases the economic burden on society [14].
For providing evidence for policy development and health care, it is essential to investigate the prevalence of dry eye during the COVID-19 pandemic.This study is the first systematic review and meta-analysis on the prevalence of dry eye during the COVID-19 pandemic.

Registration
This study was registered on PROSPERO (CRD42022369997) and conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [15].

Ethics statement
This study does not require formal ethical approval, given that all data collected in this study will not contain individual patient data, and there will be no concerns about privacy.
3. Prevalence rates can be obtained directly or calculated from relevant data.
4. The language is limited to English.

Exclusion criteria
1. Epidemiological studies conducted in eye clinics or based on dry eye patients.
2. The number of participants in the survey is smaller than 30.
3. The conference abstracts with relevant data.

Study selection and data extraction
Two researchers independently searched all databases to identify eligible studies based on inclusion and exclusion criteria and extracted data including first author, publication year, country or region, sample size, diagnostic tools, study population, age, sex composition, and prevalence.If the information was incomplete, we contacted the author(s) by e-mail.Any disagreement was resolved through discussion with a third reviewer.

Quality assessment
Two researchers independently assessed the quality of studies included in the final analysis using the Joanna Briggs Institute Quality Assessment Tool [16].According to the criteria, each item will be classified as "Yes", "No", "Unclear", and "Not applicable".Any disagreement was resolved through discussion with a third researcher.

Statistical analysis and publication bias
The prevalence of dry eye during the COVID-19 pandemic was analyzed with Stata 16 software (Stata Corp LP, College Station, TX, USA), and Odds ratio (OR) was calculated with Review Manager 5.4 (Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark).I 2 >50% was considered significantly heterogeneous.As the included studies differed in average age, study population, diagnostic tools, and time of study execution, a random-effects model were used to make our results more stable.Subgroup analysis was performed based on average age, study population, diagnostic tools, and time of study execution to detect sources of heterogeneity and to analyze the results of populations with different characteristics.The leave-one-out sensitivity analysis was performed to evaluate results and the Egger test for publication bias.

Sensitivity analysis
Leave-one-out sensitivity analysis was also conducted to examine further the possible cause of heterogeneity across the studies involved in the analysis and test the stability of the result.With this analysis, no significant change in overall prevalence and heterogeneity was caused by excluding any one study (Table 4).
The Joanna Briggs Institute Quality Assessment Tool Q1: Was the sample representative of the target population?
Q2: Were study participants recruited in an appropriate way?
Q3: Was the sample size adequate?Q4: Were the study subjects and the setting described in detail?
Q5: Was the data analysis conducted with sufficient coverage of the identified sample?Q6: Were objective, standard criteria used for the measurement of the condition?Q7: Was the condition measured reliably?
Q8: Was there appropriate statistical analysis?

Publication bias
In the current systematic review and meta-analysis, we found no evidence of potential publication bias for the prevalence of dry eye during the COVID-19 pandemic (Egger test: B = -4.92,SE = 2.766, P = 0.075).
During the COVID-19 pandemic, people's lifestyle had changed dramatically, leading to psychological disorders and affecting sleep quality [31].Not only can emotional abnormalities cause dry eye [32,33], but there is also a correlation between sleep quality and dry eye [34][35][36].For example, it has been demonstrated that lack of sleep increased tear osmolarity, reduced tear film break-up time, and lessened tear secretions, each independently triggering dry eye [35,37].At the same time, obesity and nutritional imbalance caused by home diet and prolonged wearing of masks could also cause dry eye [38,39].
It is widely believed that women's increased susceptibility to dry eye may be related to the use of contraceptive pills and hormone production in middle-aged women, which could affect the lacrimal glands, conjunctival goblet cells, lid glands, and ocular surface sensitivity, leading to dry eye [40].In the older-aged females, lower levels of estrogens and androgen may cause inadequate lacrimal gland secretion associated with aqueous deficient dry eye [41].Our results showed that the prevalence of dry eye in males was 61.2% (95%CI: 58.99%-63.6%)and 69.7% (95%CI: 54.1-85.3) in females, both are significantly higher than those in previous studies before 2019 (16.7% among females and 11.4% among males) [28].Meanwhile, our study showed a significant gender difference (OR:2.06,95%CI: 1.47-2.88,P<0.001) in the prevalence of dry eye, which is line with the results of previous studies [28,42,43].
A previous study reported an average increase of VDT by 5 hours during the pandemic in 51.1% of their study respondents [44].In addition, visual display causes blink rate and force changes, which could contribute to dry eye [45,46].Our results suggested that the OR was 2.61 (95%CI: 1.24-5.51),and there was a significant difference (P = 0.01) in the prevalence of dry eye between VDT of more than 4 hours per day (78.1%,95%CI: 71.6%-84.5%)and VDT of less than 4 hours per day (58.0%,95%CI: 43.0%-72.9%),which confirms that the prolonged use of visual display can increase the prevalence of dry eye [12-14, 28, 42, 47].
This analysis showed that the prevalence in contact lens wearers (83.5%, 95%CI: 75.7%-91.3%)was higher than in non-contact lens wearers (68.2%, 95%CI: 61.8%-71.4%),suggesting that contact lens may be a risk factor for dry eye, which is consistent with previous studies [28,42].Furthermore, our results were supported by a large epidemiologic study including office workers in Japan, revealing that contact lens wearers showed a 2.38 times higher risk of dry eye than non-contact lens wearers [48].However, there was no significant difference (P = 0.12) between contact lens wearers and non-contact lens wearers, possibly due to the lack of included studies.
The 11 studies included in our analysis were inconsistent in terms of sample size, diagnostic tools, and population characteristics; therefore, heterogeneity was considered significant, as shown by the results in Tables 3 and 4. Simultaneously, considering the differences in these studies, the results of the random-effects model, even though considered more conservative, were still used for statistical analysis to make our results more stable.
To sum up, the prevalence of dry eye during the COVID-19 pandemic is significantly higher than before, which may mean that lifestyle changes affect our eye health.In this case, protecting our eyes by changing lifestyles should be taken seriously, such as reducing visual display time, increasing outdoor activities, quitting smoking, and avoiding the wearing of contact lenses.Meanwhile, the government and medical institutions should also make more efforts to disseminate information about dry eye prevention through the Internet and media.
There are some limitations in our study.Firstly, the 11 studies included in the analysis were from 8 countries or regions in 3 continents and did not include studies performed in North America, Oceania, and Africa, leading to the possibility that the results may differ from the actual global prevalence of dry eye during the COVID-19 pandemic.Secondly, 8 of the 11 studies included in the analysis were based on school children and university students, resulting in low population representation.Finally, the 11 studies included in the analysis all used diagnostic tools, and no study using TFOS DEWS II as a diagnostic methodology, which may lead to an overestimation of the prevalence of dry eye.In addition, we found differences between diagnostic tools, which may have led to less reliable results.
Therefore, we need to include more cross-sectional surveys on the prevalence of dry eye during the COVID-19 pandemic from different countries and regions in the future.Meanwhile, studies using uniform diagnostic tools and forcing on more representative populations are also needed.

Conclusion
The prevalence of dry eye during the COVID-19 pandemic is significantly higher than before, and a higher prevalence is found among females and those having a visual display time of more than 4 hours per day.

Fig 2 .
Fig 2. Forest plot of studies showing the overall prevalence of dry eye during the COVID-19 pandemic.The random-effects model was used.The pooled estimate was reported as prevalence as a percentage and CI.CI, Confidence interval.https://doi.org/10.1371/journal.pone.0288523.g002

Fig 3 .
Fig 3. Forest plot of studies showing the prevalence of dry eye during the COVID-19 pandemic in Asia.The random-effects model was used.The pooled estimate was reported as prevalence as a percentage and CI.CI, Confidence interval.https://doi.org/10.1371/journal.pone.0288523.g003

Fig 4 .Fig 5 .Fig 6 .
Fig 4. Forest plot of studies showing the prevalence of dry eye in females and males.The random-effects model was used.A and B refer to the prevalence of dry eye in females and males during the COVID-19 pandemic, respectively.The pooled estimate was reported as prevalence as a percentage and CI.CI, Confidence interval.https://doi.org/10.1371/journal.pone.0288523.g004

Fig 7 .Fig 8 .
Fig 7. Forest plot of studies comparing the prevalence of dry eye between VDT of more than and less than 4 hours per day.The random-effects model was used.The pooled estimate was reported as an odds ratio.VDT, visual display time; CI, Confidence interval; IV, Inverse variance.https://doi.org/10.1371/journal.pone.0288523.g007