Association Between Depression and Severity of Dry Eye Symptoms, Signs, and Inflammatory Markers in the DREAM Study

Key Points

Question  Is severity of dry eye symptoms and signs associated with presence of depression?

Findings  In this multicenter study of 535 participants with dry eye disease, those who screened positive for depression had worse dry eye symptoms and overall dry eye signs but similar inflammatory markers compared with those of participants with dry eye who screened negative for depression.

Meaning  In this study, patients with dry eye disease and depression may have more severe dry eye symptoms and signs than those without depression.

Importance  Depression is more prevalent in patients with dry eye disease (DED) than in the general population; however, the association between severity of DED and depression needs further evaluation.

Objective  To investigate the association between depression and severity of DED symptoms and signs, including inflammatory markers.

Design, Setting, and Participants  Secondary cross-sectional and longitudinal analysis performed in April to December 2020 of data from Dry Eye Assessment and Management (DREAM) study, a randomized clinical trial from October 2014 to July 2016 including patients with moderate to severe symptoms and signs of DED. Enrolled from 27 ophthalmology and optometry centers, both academic and private, in 17 US states, 535 patients were followed up for 1 year.

Exposure  Participants screened positive for depression if they scored 42 or less on the Mental Component Summary (MCS) of the 36-Item Short Form Health Survey.

Main Outcomes and Measures  Symptoms of DED were assessed by Ocular Surface Disease Index (OSDI) and Brief Ocular Discomfort Index (BODI) and signs assessed by tear film breakup time, Schirmer test, corneal and conjunctival staining, tear osmolarity, and meibomian gland dysfunction at baseline, 6 months, and 12 months. A composite severity sign score was calculated from all 6 signs. Inflammatory markers (cytokines in tears and HLA-DR expression by conjunctival surface cells) were measured for some trial participants. Features of DED were compared between participants with and without depression and adjusted for age, sex, race, visits, and baseline comorbidities.

Results  Among the 535 participants, mean (SD) age was 58 (13.2) years, 434 participants (81%) were women, and 398 (74.4%) were White. Participants who screened positive for depression had worse DED symptoms by OSDI (effect size = 0.45, P < .001) and BODI (effect size = 0.46, P < .001) and composite DED sign score (effect size = 0.21, P = .006). Lower MCS score (ie, worse depression) was correlated with higher OSDI score (ie, worse DED symptoms) at baseline (Spearman ρ = −0.09, P = .03), 6 months (ρ = −0.20, P < .001), and 12 months (ρ = −0.21, P < .001). Inflammatory markers did not differ by depression status.

Conclusions and Relevance  Depression was associated with more severe dry eye symptoms and overall signs, suggesting that among patients with moderate to severe DED, those with depression may be likely to have more severe DED. These findings support consideration of depression as a comorbidity when managing patients with DED. Further study is needed to elucidate the relationship.

Dry eye disease (DED) is a common inflammatory condition that interferes with quality of life through ocular pain and irritative symptoms.¹ Patients with DED are more likely to report trouble with daily tasks such as reading and driving, which can affect general and emotional well-being.^2,3 Depression is more prevalent among patients with DED, although the underlying mechanisms are not well understood.⁴ Patients with DED may be more at risk for depression involving pathways through decreased quality of life, shared genetic factors, changes to central pain processing, and anticholinergic effects of antidepressant medications.^4,5 Inflammation has also been implicated in the pathogenesis of both DED⁶ and depression,⁷ but little is known about whether the same inflammatory processes affect both conditions. Given that DED and depression are both prevalent conditions that affect daily well-being, studies are needed to better understand their association.

Prior studies found that depression was associated with increased risk of DED,^4,8 but association of depression with severity of DED symptoms and signs has not been comprehensively studied. A few single-center cross-sectional studies found depression scores to be correlated with symptoms but not individual signs of DED and did not explore the role of inflammation.^9-12

To further elucidate the association between depression and severity of DED as well as the potential role of inflammation in both, we performed secondary analyses of data from the 12-month, multicenter Dry Eye Assessment and Management (DREAM) study.¹³

Details of the DREAM study have been previously described, and the trial is registered with ClinicalTrials.gov (NCT02128763). Only major features of DREAM relevant to this analysis are described below.^13,14 Our reporting of this analysis follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.¹⁵

The DREAM study was a multicenter, randomized clinical trial to evaluate the efficacy of ω-3 fatty acid supplements compared with placebo supplements, with score of Ocular Surface Disease Index (OSDI) as primary outcome.¹³ Between October 2014 and July 2016, 535 participants were enrolled from 27 clinical centers in 17 states of the United States. Participants were randomized 2:1 to receive ω-3 supplements or placebo. Participants signed informed consent forms. The study received institutional review board approval for each center and adhered to the tenets of the Declaration of Helsinki. Sex and race were reported by participants; choices for race included American Indian or Alaskan Native, Asian, Black or African American, and White.

The eligibility criteria for the study included age 18 years or older, DED symptoms for at least 6 months, use of artificial tears at least twice a day for 2 weeks before screening, and OSDI score of 25 to 80 at a screening visit and 21 to 80 at a baseline eligibility-confirmation visit.¹⁴ Participants had at least 2 of the 4 following signs of DED in the same eye at both screening and baseline visits: conjunctival staining score 1 or greater, corneal fluorescein staining score 4 or greater, tear film breakup time (TBUT) 7 seconds or less, and Schirmer test score of 1 to 7 mm per 5 minutes. Participants had evaluations of DED symptoms, signs, and inflammatory markers and depression at baseline and 6-month and 12-month follow-up visits.

Severity of DED symptoms was evaluated primarily by the widely used OSDI.¹⁶ Ocular Surface Disease Index scores range from 0 to 100, with higher scores indicating more severe DED symptoms. Severity of ocular discomfort was also measured by the third item from the Brief Ocular Discomfort Inventory (BODI),¹⁷ with scores of 0 to 100 (higher score indicating greater discomfort).¹⁸

Signs of DED were evaluated in each eye by corneal fluorescein staining (range 0-15, with 15 indicating greatest abnormality),¹⁹ TBUT (lower scores indicating greater abnormality), Schirmer test (lower scores indicating greater abnormality), conjunctival lissamine green staining (range 0-6, with 6 indicating greatest abnormality),¹⁹ and tear osmolarity (higher scores indicating greater abnormality). Meibomian gland dysfunction was assessed by clinicians based on degree of meibomian gland plugging and secretion at the lower eyelid margin (range 0-6, higher scores indicating more severe dysfunction).²⁰

Conjunctival impression cytology samples were collected from 1049 eyes of 527 participants. Samples were assayed for percentage HLA-DR positive in total cells, epithelial cells, and white blood cells. Detailed methodology for evaluation of HLA-DR expression has been published.²¹

Tear samples were collected for evaluating proinflammatory cytokines from 218 participants from 10 centers that had the required freezer storage capability. Sufficient tear volume (pooled from 2 eyes) was available from 131 participants (57.0%) for analysis of proinflammatory cytokines, including interleukin 1β (IL-1β), IL-6, IL-8, IL-17A, IL-10, interferon γ (INF-γ), and tumor necrosis factor (TNF-α).²² Detailed methods of tear sample processing and assaying have been published.²³

The Medical Outcomes Study 36-Item Short Form Health Survey (SF-36) version 2.0 was administered to each participant. The SF-36 is widely used for assessing patient-perceived well-being in multiple domains.^24,25 Two summary SF-36 scores are generated: the Physical Component Summary (PCS) and Mental Component Summary (MCS). Scores for MCS range from 0 to 100, with higher scores indicating greater psychological well-being. The MCS has been used for depression screening in various patients.^25-29

Our primary measurement of depression was predefined as SF-36 MCS score of 42 or lower. This cutoff was based on a study of the US general population comparing SF-36 with 2 other established depression evaluations: Center for Epidemiological Studies–Depression (CES-D) questionnaire and diagnostic interview following the National Institute of Mental Health Diagnostic Interview Schedule criteria.^24,25 The recommended cutoff of 42 has sensitivity of 73.7% and specificity of 80.6% for identifying clinical depression.²⁴

We examined 2 secondary measures of depression via self-report of depression and use of antidepressant medications (eTable 1 in the Supplement). At enrollment, clinical coordinators asked participants about their medical history of depression in the past 2 years (none, positive past history, ongoing) to determine self-reported depression and used a medication log to collect information on current medications, reason for taking, and starting and ending dates to determine antidepressant-defined depression.

We evaluated the association between SF-36 MCS score and DED symptoms, signs, and inflammatory markers using 2 approaches. The first approach used MCS score to define presence of depression (MCS score ≤42) and compared the severity score of DED indicators (symptoms, signs, and markers) between participants with vs without depression. The second approach used MCS scores as continuous measures of severity of depressive symptoms and assessed the association with the severity scores of DED indicators with Spearman correlation coefficients (ρ).

We performed additional analyses using depression defined by self-report and by participant use of antidepressant medications. For the comparison of each DED sign between participants with vs without depression, the sign from the worse eye of the specific sign at the same time point was used. In addition, we adapted a method from previous studies to calculate a composite severity score based on 6 DED signs (TBUT, Schirmer testing, corneal and conjunctival staining, tear osmolarity, and meibomian gland dysfunction). Composite scores range from 0 to 1, with 1 indicating the most severe DED signs.^30-32

All these comparisons were made using generalized linear models without and with adjustment for age, sex, race, and time of visit (ie, baseline, 6 months, or 12 months). We also adjusted by several self-reported comorbidities, including smoking, rosacea, rheumatoid arthritis, peripheral artery disease, and Sjögren syndrome,³³ which have been found to be associated with severity of DED in the DREAM study.³⁴

We performed these analyses using the combined data from baseline, 6 months, and 12 months to improve statistical power, and the correlation of repeated measures was accounted for using generalized estimating equations. Similar analyses were performed for each time point separately to check the consistency of results over time. Effect sizes were calculated by [mean difference]/[SD] to assess the magnitude of associations. Because the DREAM study did not find a significant effect of ω-3 supplementation on DED signs and symptoms compared with placebo,¹³ evaluations were based on the data from the 2 treatment groups combined. In addition, ω-3 has not been found to have any protective effect on depression.³⁵

Because of the skewed distributions of inflammatory markers, comparisons between participants with vs without depression were made using the Wilcoxon rank sum test for each time point and the clustered Wilcoxon rank sum test for all time points combined.³⁶

Statistical analyses were performed using SAS version 9.4 (SAS Institute). P values were 2-sided without adjustment for multiple comparisons.

Among the 535 DREAM participants, mean (SD) age was 58 (13.2) years, 434 (81%) were women, and 398 (74.4%) were White (Table 1). For their follow-up, 479 participants (89.5%) completed the 6-month visit and 486 (90.7%) completed the 12-month visit.

Among all participants, 84 (15.7%) had depression (≤42 on MCS) at baseline, with 82 (17.3%) at 6 months and 64 (13.2%) at 12 months (P = .32 and P = .09 compared with baseline, respectively). Mean (SD) MCS score was 52.3 (9.4) at baseline, 51.7 (9.8) at 6 months, and 52.2 (9.0) at 12 months (P = .15). The ω-3 and placebo groups had similar mean MCS score and proportion with depression at all time points (eTable 2 in the Supplement).

Participants who screened positive for depression were similar in age, sex, and race to participants who screened negative for depression (Table 1). However, those who screened positive for depression reported higher prevalence of having rheumatoid arthritis (difference, 9.3%; 95% CI, 1.0%-17.7%; P = .007), lower prevalence of having Sjogren syndrome (difference, −5.9%; 95% CI, −11.3% to −0.5%; P = .048), and more current or former smoking (difference, 5.5%; 95% CI, −1.0% to 12.1%; and 6.6%; 95% CI, −4.1% to 17.4%; respectively; P = .03).

At baseline, 87 participants (16.3%) self-reported ongoing depression and 35 (6.5%) reported history of depression while 118 participants (22.1%) reported taking antidepressants. More participants who screened positive for depression vs those who screened negative for depression reported ongoing (44.0% vs 11.1%) or past depression (17.9% vs 4.4%) (P < .001) or taking antidepressants (44.0% vs 18.0%, P < .001, Table 1).

Lower MCS scores (more severe depression) were correlated with higher OSDI scores (more severe DED symptoms) at baseline (ρ = −0.09, P = .03), 6 months (ρ = −0.20, P < .001), and 12 months (ρ = −0.21, P < .001) (eTable 3 in the Supplement). However, change in MCS score from baseline was not associated with change in OSDI score at 6 months (ρ = −0.03, P = .52) or 12 months (ρ = −0.04, P = .36) (eTable 4 in the Supplement).

In analysis of combined data from all time points, participants who screened positive for depression had a higher mean (SD) OSDI score (42.2 [18.8] vs 33.9 [18.1]) and more severe ocular discomfort on the BODI (45.5 [21.5] vs 35.7 [21.0]) than participants who screened negative for depression (P < .001).

These differences remained after adjustment by age, sex, race, and visits and with further adjustment by comorbidities (Table 2). The effect size for OSDI and ocular discomfort on BODI were 0.45 and 0.46, respectively (Table 2). Similar associations between positive depression and DED symptoms were found at baseline (Table 3), 6 months (eTable 5 in the Supplement), and 12 months (eTable 6 in the Supplement).

Mean OSDI score was not different between participants taking vs not taking antidepressants (35.7 [17.5] vs 34.1 [18.8]; adjusted P = .39; eTable 7 in the Supplement) or between participants with vs without self-reported ongoing or a history of depression (36.2 [17.0] vs 34.0 [18.9]; adjusted P = .21; eTable 8 in the Supplement).

Mental Component Summary score was not correlated with any of the 6 DED signs or composite severity score of DED signs at baseline, 6 months, or 12 months (all ρ ≤ 0.08, P ≥ .07; eTable 3 in the Supplement). However, at 6 months, a positive change in MCS score from baseline (improvement) correlated with a negative change of tear osmolarity (improvement) from baseline (ρ = −0.11, P = .03) and a negative change of composite severity score of DED signs (improvement) from baseline (ρ = −0.10, P = .03; eTable 4 in the Supplement). At 12 months, a positive change in MCS score from baseline (improvement) was positively correlated with change in Schirmer test (improvement: ρ = 0.09, P = .04; eTable 4 in the Supplement).

Participants who screened positive for depression had a higher mean composite severity score for DED signs than participants who screened negative for depression when all time points were combined (mean [SD]: 0.55 [0.28] vs 0.49 [0.29]; effect size = 0.21; adjusted P = .01; Table 2), at baseline (0.57 [0.25] vs 0.49 [0.29]; adjusted P = .01; Table 3), and at 6 months (0.54 [0.30] vs 0.49 [0.29]; adjusted P = .04; eTable 5 in the Supplement) but not at 12 months (0.53 [0.27] vs 0.50 [0.29]; adjusted P = .60; eTable 6 in the Supplement). The mean corneal staining score was higher in the group who screened positive for depression than the group who screened negative when all visits were combined (4.52 [3.01] vs 3.97 [3.10]; effect size = 0.18; adjusted P = .03; Table 2) and at 6 months (4.64 [3.11] vs 3.79 [3.02]; adjusted P = .006; eTable 5 in the Supplement).

Participants who screened positive for depression at baseline had more improvement at 12 months in conjunctival staining than participants who screened negative for depression (−0.92 [1.28] vs −0.43 [1.30]; adjusted P = .003) but had no differences in other DED signs or differences at 6 months (eTable 9 in the Supplement).

Dry eye disease signs were not different according to use of antidepressant medications (eTable 7 in the Supplement) or self-reported depression (eTable 8 in the Supplement).

Participants who screened positive for depression did not have different HLA-DR percentages in total cells, epithelial cells, or white blood cells compared with participants without depression when all time points were combined (eTable 10 in the Supplement), but had lower HLA-DR percentages in total cells at 12 months (median = 5.4 vs 7.4; P = .02; eTable 6 in the Supplement). No differences in tear cytokines were observed between the participants who screened positive or negative for depression, either when all time points were combined (eTable 10 in the Supplement) or at individual time points (eTables 3, 5, and 6 in the Supplement).

Score on the MCS was associated with HLA-DR percentage in total cells (ρ = 0.13, P = .005) and in epithelial cells (ρ = 0.10, P = .04) at 12 months, but not at baseline or 6 months (eTable 3 in the Supplement). No associations were found between MCS score and tear cytokines at baseline, 6 months, or 12 months (eTable 3 in the Supplement). However, a positive change in MCS score (improvement) from baseline was correlated with a negative change in IL-1β at 6 months (ρ = −0.23, P = .048) and in IL-6 (ρ = −0.29, P = .02) and IL-8 (ρ = −0.25, P = .04) at 12 months (eTable 4 in theSupplement).

Participants taking antidepressants had higher IL-6 at 6 months (median = 6.78 vs 3.74; P = .03; eTable 7B in the Supplement) but not in other cytokines (eTables 7, 7A, 7B, and 7C in the Supplement) than participants not taking antidepressants. Participants who reported history of depression had lower HLA-DR percentages in total cells at baseline (median = 6.63 vs 9.67; P = .04; eTable 8A in the Supplement) and at 12 months (median = 6.20 vs 7.46; P = .03; eTable 8C in the Supplement), lower HLA-DR percentage in epithelial cells (median = 5.19 vs 8.00; P = .005), and higher TNF-α at baseline (median = 1.47 vs 0.00; P = .03) than participants without history of depression (eTable 8A in the Supplement).

Our study found that among participants with moderate to severe DED, those who screened positive for depression had worse DED symptoms, more ocular discomfort, worse corneal staining scores, and worse composite severity scores of DED signs compared with participants who screened negative for depression. Our study also evaluated inflammatory markers but did not find consistent evidence for the role of inflammatory factors in accounting for observed association between depression and DED severity.

Several previous studies reported participants with depression had worse DED symptoms but did not find significant associations with individual DED signs.^9-12 Our participants with DED who screened positive for depression had both more severe DED symptoms and overall signs than participants who screened negative for depression. Discomfort of DED and interference with daily activities contribute to lower quality of life and may put patients at greater risk for depression.^2,37,38 Conversely, patients with depression were found to spend more time watching TV and using a computer, and increased screen time may contribute to DED.³⁹ Others postulated that depression may be associated with DED symptoms because of changes in pain sensitization in patients with depression or somatic manifestations of depression.^10,12 Our participants who screened positive for depression also had worse mean composite DED sign scores (although with smaller effect size than DED symptoms) and corneal staining scores, suggesting DED severity of signs, rather than patient perception alone, may have a role in an association between DED and depression.

We examined several tear inflammatory markers that have been implicated in pathogenesis of both DED²² and depression.⁴⁰ A previous study found levels of IL-6, IL-17, and TNF-α were higher in participants with depression with DED compared with controls, but it is unclear whether this was due to depression or DED.⁴¹ In our participants with moderate to severe DED, inflammatory cytokine levels were not different between participants who screened positive vs negative for depression and were not correlated with severity of depressive symptoms. Pattern of inflammatory marker expression in tears did not appear to be affected by depression status in participants with DED.

Previous studies have found antidepressants to be a risk factor for DED, and antidepressant use may indicate a patient’s potential depression status.^13,42,43 However, we did not find use of antidepressants to be associated with more severe DED symptoms or signs. We also did not find an association between self-reported depression and DED symptoms or signs.

Generalizability of our findings is enhanced by enrolling a large number participants throughout the United States with 1 year of follow-up. Unlike studies that only examined association between depression and individual signs, we used a composite sign score. Because there is no single gold standard metric for measuring dry eye severity, the composite score provides a more comprehensive measure of dry eye signs.^30-32 Furthermore, we explored the potential role of inflammation linking depression and DED and found that inflammatory markers in tears were not associated with depression.

Our study had limitations. Use of a questionnaire may result in misclassifying depression, diluting the strength of associations. However, our prespecified screening cutoff (≤42 in MCS score) has been validated in the US general population and used in various clinical trials to screen for depression.^20-25 In addition, participants may have been taking medications classified as antidepressants for indications other than clinical depression. Also, the composite severity score may be driven by corneal staining score. However, TBUT, Schirmer test, meibomian gland dysfunction, and tear osmolarity all trended in the same direction, suggesting composite sign score may capture small differences of each DED that contribute to overall severity of measurable DED signs.^30-32 Although we did not adjust for multiple comparisons, we checked results consistency across each time point (baseline, 6 months, and 12 months) to form conclusions. One limitation with tear cytokine assessment was that participants who provided sufficient tear samples were younger and had less severe symptoms and signs than participants without sufficient tear samples. In addition, the DREAM study enrolled only participants with symptoms and signs of moderate to severe DED, and the absence of patients with mild DED may have limited the ability to detect an association with depression.

Symptoms and overall signs of DED were more severe in participants with moderate to severe DED who screened positive for depression, suggesting depression may be associated with more severe DED. Identifying depression and considering treatment, including systemic medications, may be useful in managing patients with DED. Patients with more severe DED concerns or sign measurements may benefit from comorbid psychiatric screening. The cause of the observed association needs further investigation.

Accepted for Publication: January 17, 2022.

Published Online: March 10, 2022. doi:10.1001/jamaophthalmol.2022.0140

Corresponding Author: Gui-shuang Ying, PhD, 3711 Market St, Ste 801, Philadelphia, PA 19104 (gsying@pennmedicine.upenn.edu).

Author Contributions: Dr Ying had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Zhou, Asbell.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Zhou, Asbell.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Yu.

Obtained funding: Asbell, Maguire, Ying.

Administrative, technical, or material support: Zhou, Asbell.

Supervision: Sayegh, Asbell, Maguire, Ying.

Conflict of Interest Disclosures: Dr Murrough reported having provided consultation services and/or having served on advisory boards for Allergan, Boehringer Ingelheim, Clexio Biosciences, Fortress Biotech, FSV7, Global Medical Education (GME), Novartis, Otsuka, Sage Therapeutics, and Engrail Therapeutics outside the submitted work. Dr Sayegh reported being a consultant for Novartis and Allergan during the conduct of the study. Dr Asbell reported being a consultant for Topovert, Sanofi, Blephe, Senju, Axero, Eyepoint, Shire, Novaliq, Kala, Sun, Dompe, Santen, Alcon, Allakos, Regeneron, and ECLA outside the submitted work; reported grants from the National Institutes of Health (NIH), National Eye Institute (NEI), and NIH Office of Dietary Supplements during the conduct of the study; and reported grants from Regeneron, Allakos, Miotech, MC2, Sylentis, and Research to Prevent Blindness outside the submitted work. Dr Maguire reported having received personal payments for service on data and safety monitoring committees for Genentech and reported grants from the NEI during the conduct of the study. Dr Ying reported being a biostatistical consultant for Ziemer Ophthalmic Systems AG and Synergy Research Inc and reported grants from the NEI during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was supported by NIH grants U10EY022879, U10EY022881, and R21EY031338.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The DREAM Study Research Group: The members of the DREAM Study Research Group are listed in Supplement 2.

Meeting Presentation: This work was presented as a poster at the Annual Meeting of the Association for Research in Vision and Ophthalmology; May 7, 2021; virtual.