The present study has several limitations. First, history regarding systemic medication was not investigated. Systemic medications including antidepressants, antianxiety medications, antihistamines, diuretics, and antibenign prostate hypertrophy medications were reported to be associated with DED.
14,15 However, no significant correlation was found between Schirmer and SGDS-K scores (
Table 1). Moreover, the association between DED and depression was significant only in the high Schirmer score group; this suggests that the influence of systemic medications was not critical (
Fig. 1). However, we believe that the thorough investigation of systemic medication history is necessary in further studies to rule out possible compounding effects.
14 Second, 62.0% (657/1060) of the subjects enrolled in the study sample participated in the initial survey, and only 21.2% (139/657) of them underwent secondary ophthalmologic examinations. This fact may raise a concern about selection bias, although we used the following measures to avoid bias: in the initial survey, specific statements about DED or depression were never included in the invitation letter; all 657 subjects who finished the initial survey were invited for the second step of the study; and in the second step, the diagnostic status of DED was masked to the examiner as well as to all participants. However, multivariate analyses revealed the association between the two diseases (
Table 2). The SGDS-K score correlated with the number of positive responses in the DEQ (
Table 1), suggesting the correlation between severity of the depressive mood and that of DED symptoms. Moreover, the DED symptoms and depression were evaluated in the initial survey, thus the possible bias due to the small number of participants in the second step had little influence on the determination of association between symptom-based DED and depression. Therefore, we believe that the results demonstrate the association between the two diseases, regardless of the bias. Third, because of the cross-sectional nature, this study could not show a causal relationship. It remains unclear whether depression is a predisposing factor of DED, or vice versa, although we believe that both conditions have an influence on each other. Fourth, the severity of dry eye symptoms was represented by the number of positive responses to DEQ. Subjective symptoms may be better quantified using tools including the visual analog scale and the ocular surface disease index score.
35,36 However, considering that participants were elders who often have difficulty answering excessive numbers of questions, we evaluated the symptom grade indirectly rather than adding another questionnaire. Fifth, our diagnostic criteria of DED have still limited value. As the lack of association between DED signs and symptoms is well reported,
37 –41 we selected the symptom-based diagnostic criteria as did the recent population-based studies.
9,17,39,42,43 However, our DEQ may not be specific for DED and also be reflective of other ocular surface diseases including MGD, allergic and chronic infectious conjunctivitis.
9 Moreover, some dry eye symptoms might not be included in our DEQ.
9 Therefore, the development of more accurate diagnostic criteria for DED is necessary.
9 Finally, we did not control for contact lens wear, which is an important risk factor of DED. However, this study included only subjects aged 65 years or older, among which the number of contact lens wearer could be negligible in Korea.