Abstract

INTRODUCTION

Dry eye is a multifactorial disease characterized by ocular surface changes which result in tear film impairment, eye itchiness, redness, light hypersensitivity, blurred vision, and other symptoms and/or discomforts[1]–[2]. Dry eye is estimated to affect 5%-35% of worldwide population at various ages with increasing frequency[3]. In the United States, moderate and/or severe dry eye affected more than 3.2 million of female and 1.6 million of male population at the age of above 50 years old[4]–[5]. In China, the incidence of dry eye is also increasing because of air pollution, the popularization of computer and mobile phone, and rapid aging of the population, which is becoming an increasingly significant public health problem.

Dry eye is associated with many risk factors such as environment, lifestyle, age, sex, drug history, and systemic diseases, among which the lifestyle factors may play an important role[3]. Smoking is already known as an important risk factor for many chronic diseases and however is still an unclear risk factor of dry eye[3]. A number of studies evaluated the association between smoking and dry eye risk[6]–[19], which show that smoking could increase the risk of dry eye[12]–[13],[17]–[18]. However, other researches showed disputed results[7]–[8],[10]–[11],[14]–[15],[19]. The definition of dry eye was suggested according to tear break-up time (TBUT) and Schirmer score[2], and smoking was found to decrease TBUT in some of the studies[6],[9],[16], in which Sayin et al[9] found the decreasing of Schirmer score in smokers. Among the analyses, most of them showed their investigations were performed in specific populations. However, these results may not be representative of the larger population. Therefore, this Meta-analysis aims to summarize the association between smoking and the risk of dry eye, and to provide useful evidence for this association.

MATERIALS AND METHODS

RESULTS

Ten studies[7]–[8],[10]–[14],[17]–[18],[21] which fulfilled the inclusion criteria were identified (refer to the flow diagram in Figure 1), including two cohort studies[10],[13] and eight cross-sectional surveys[7]–[8],[11]–[12],[14],[17]–[18],[21]. Two cohort studies[10],[13] were both from the United States. Three of the eight cross-sectional surveys included[17]–[18],[21] were from China, two[7],[14] from Japan, one[8] from India, one[11] from Australia, and one[12] from Indonesia. The study samples included in the Meta-analysis were summarized in Table 3.

DISCUSSION

In our primary Meta-analysis, the effect of smoking on dry dye was evaluated. Our results showed no significant association between current smoking and increased risk of dry eye, but the P value was close to 0.05 (P=0.055). We also found no significant association in the analyses on ever smoking (P=0.107).

Several pathogenesis of dry eye including chronic inflammation of the ocular surface, decreased sensitivity of cornea and conjunctiva, reduction of production and/or stability of tears, and epithelial damage have been suggested[1]–[2]. As mentioned, smoking is a well-known risk factor for many chronic diseases, which also affects the eyes. The health of the eye could be affected by smoking toxins which decrease blood flow and/or accelerate thrombus formation in ocular capillaries[22]. Cigarette smoking may also cause a higher risk of many eye diseases such as cataracts, age-related macular degeneration, glaucoma, diabetic retinopathy, and optic neuritis[22], but the relationship between smoking and the risk of dry eye was still unclear. The pooled analysis by Thomas et al[16]. and some other studies[6],[9] found that the tear stability and the sensitivity of cornea and conjunctiva could be decreased by smoking which also showed the decreased TBUT. Also, for dry eye patients, smoking had been reported to induce discomforts including burning and foreign body sensation of the eyes[23]–[24]. But for the results of Schirmer score, it showed no significant change[6],[16]. On the International Dry Eye Workshop 2007, the diagnostic criteria of dry eye were suggested to be TBUT ≤10s and Schirmer score performed without anesthesia ≤5 mm in 5min[2]. The possible explanation is that there were no clear diagnostic criteria of dry eye in the studies or just using questionnaire. It may affect the association between dry eye and smoking of some studies, thus cause the bias.

Dry eye is associated with many risk factors such as environment as mentioned. Among the risk factor of dry eye, older age and female sex were two of the most aconsistent ones[3],[25]. The effect of sex hormones on the homeostasis of ocular surface had been acknowledged. It was reported by Sullivan et al. that androgen levels decrease with advancing age in both males and females[26]. Clinically, patients suffering from sex hormone deficiency including congenital androgen insufficiency syndrome[27]–[28], Sjogren's syndrome[29], premature ovarian failure[30], and those receiving anti-androgen medications[31]–[33] usually had higher risk of dry eye. Also, it is known that smoking is more prevalent in male population than female. Five studies in which adjusted age and gender were included in the analysis[7],[11]–[14]. When age and gender were adjusted, it also showed no statistically significant relationship between current smoking and dry eye (OR=1.16; 95% CI: 0.83-1.64; P=0.383).

Apart from age and sex, environment may also affect the incidence of dry eye. For example, the office workers who use visual display terminals (VDTs)[34] and glaucoma patient who use anti-glaucoma medication[35] were found to have higher risk of dry eye. In this Meta-analysis, the result of no significant association was changed between current smoking and dry eye when only the studies which sample is general population were included. The possible explanation for the inconsistency of the result is that some of the studies included in our primary Meta-analysis were related to the populations exposed to other risk factors of dry eye such as the use of VDTs, or the studies were related to specific population such as hospital-based population.

Several strengths and limitations of this Meta-analysis should be understood for proper interpretation of the results and findings. For the strengths, the first is this Meta-analysis included the studies in several important English databases published from January 1964 to December 2015. Second, the adjustment of age and gender and the subgroup analysis of general population were performed to avoid the bias of different age and sex, and to reduce the influence of different population including the patients of other eye diseases and/or visual display terminal users, and positive results was found in the analysis of general population. The most important is that the result of this Meta-analysis may be helpful to the ophthalmologists' suggestion of lifestyle on dry eye patients. For the limitations, the first is that most of the articles included are cross-sectional surveys which may affect the quality of this analysis. Second, our results were probably be affected by misclassification of smoking, and we did not include the studies related to passive smoking. And our results may be affected by different factors although we did sensitivity analysis of general populations and adjusted age and gender. Third, among the studies included, some of them showed different or unclear diagnostic criteria and/or types of dry eye, which may also affect our results. The statistically significant heterogeneity showed by our result could not be avoided. In addition, present data were inadequate to carry out dose-dependent risk estimation.

In summary, this Meta-analysis of all ten studies shows negative association between current and/or ever smoking and increasing risk of dry eye. But when only the studies which sample is general population were included, it showed statistically significant relationship between current smoking and dry eye. The findings would be possible to indicate that smoking may associate with the risk of dry eye in general population which may be helpful to the ophthalmologists' suggestion on dry eye patients. This Meta-analysis cannot explore the association between different types of dry eye and smoking due to the limitation of data in this study. Therefore, further investigations and Meta-analysis are needed to validate the role for smoking in the incidence of dry eye.

Acknowledgments

REFERENCES