Prevalence of dry eye disease and its association with sleep quality and depression: a hospital-based survey in Thai population

Abstract

Objectives

To estimate the prevalence of dry eye disease (DED) and explore its association with depression and poor sleep quality.

Design

A cross-sectional study.

Setting

The study was conducted at the ophthalmology outpatient clinic of a tertiary university hospital in Thailand, from September 2022 to April 2023.

Participants

A total of 1321 patients aged 18 years or older, without any history of orbital disease, active superficial or intraocular infection/inflammation, eyelid pathology, or prior intraocular or eyelid surgery within the past 6 months, were enrolled in the study.

Interventions

All patients underwent dry eye examination, including the Ocular Surface Disease Index questionnaire, tear break-up time and ocular surface staining. Physical activity was measured using the Global Physical Activity Questionnaire, which was expressed as total Metabolic Equivalent of Task-minutes per week. Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI), Composite Scale of Morningness (CSM) and STOP-Bang. Depressive symptoms were measured using the Patient Health Questionnaire (PHQ-9).

Primary and secondary outcome measures

Prevalence of DED and association pathways involving depression, sleep quality and DED.

Results

The mean age of the participants was 58.3±14.0 years, with a female predominance of 73.7%. Of the 1321 patients, 668 were diagnosed with DED, resulting in a hospital-based prevalence of 50.6% (95% CI: 47.8%, 53.3%). After adjusting for age, sex and underlying disease, PSQI, PHQ-9, STOP-Bang and CSM scores remained significantly associated with DED with adjusted ORs of 2.48 (95% CI: 1.96, 3.14; p<0.001), 1.65 (95% CI: 1.05, 2.61; p=0.031), 1.81 (95% CI: 1.05, 3.14; p=0.033) and 1.32 (95% CI: 1.04, 1.68; p=0.023), respectively. The effects of depression on DED were mediated indirectly via poor sleep quality (OR=1.32; 95% CI: 1.18, 1.49; p<0.001) and directly via other mechanisms (OR=1.66; 95% CI: 1.06, 2.58; p=0.021).

Conclusions

DED is notably common among Thai patients. Depression and poor sleep quality are significantly associated with DED. Poor sleep quality may mediate the relationship between depression and DED.

Strengths and limitations of this study.

• This large cross-sectional study, conducted in the Southeast Asia region, specifically focuses on the associations between dry eye disease, depression and poor sleep quality.

• The study incorporated both subjective and objective assessments of dry eye disease to provide a well-rounded analysis of patients.

• An additional mediation analysis was performed to investigate whether the effect of depression on dry eye disease might be mediated through sleep quality.

• The study population consisted of patients who attended an outpatient eye clinic of a tertiary hospital centre, thereby limiting the generalisability of the findings to a broader general population.

• A causal relationship could not be definitively established due to the nature of the cross-sectional study design.

Introduction

Dry eye disease (DED) is a chronic, multifactorial, ocular surface disease characterised by an instability of tear film homeostasis that causes undesirable ocular symptoms such as stinging, burning or a scratchy sensation in the eyes, visual fluctuation, sensitivity to light and red eyes.¹ This condition significantly affects patients’ quality of life and has become a common ophthalmic concern worldwide. The prevalence of DED varies widely, ranging from 5% to 74% across different populations.^{2 3} Numerous risk factors for DED include intrinsic factors (such as age, sex, underlying diseases, psychiatric status, medication use and previous eyelid or ocular surgeries), lifestyle behaviours (such as contact lens use, use of cosmetics, digital screen use, dietary habits, sleep and daytime activities) as well as environmental factors (such as low humidity, extreme temperature and air pollution).^{3 4}

Recent evidence has reported significant associations between DED, depression and poor sleep quality.5,7 Additionally, among patients with moderate to severe DED, those with depression are likely to have more severe symptoms and overall signs.⁸ However, the underlying causal pathways remain uncertain. It is widely recognised that depression directly affects sleep duration and sleep quality and contributes to various somatic symptoms. Inadequate sleep is also linked to autonomic, hormonal and neurochemical alterations that influence increased norepinephrine and stress hormone levels, which result in elevated blood pressure while reducing parasympathetic tone.⁹ Since the parasympathetic nervous system regulates tear secretion from the lacrimal glands, poor sleep quality may subsequently reduce tear secretion and lead to DED.⁹

We prospectively conducted a cross-sectional study with the following aims. First, to estimate the prevalence of DED and identify factors associated with DED in a hospital-based population. Second, to assess whether the effect of depression on DED is mediated through sleep quality. Our findings reflect the importance of considering a holistic approach that evaluates both sleep quality and depression symptoms in patients with DED.

Materials and methods

This study was a cross-sectional design at the outpatient clinic of the Ophthalmology Department, Ramathibodi Hospital, Bangkok, Thailand, from September 2022 to April 2023.

Participants

All potential participants were selected from patients or their relatives/companions on the waiting list for a consultation at the outpatient ophthalmology clinic. Participants were eligible if they met the following inclusion criteria: (1) aged 18 years or older, (2) demonstrated verbal communication and (3) was willing to participate and provide informed consent. Patients with the following conditions were excluded: any form of orbital disease, active superficial or intraocular infection/inflammation, eyelid pathology, or previous intraocular or lid surgery within the past 6 months.

Patient and public involvement

Patients and the public were not involved in the development, design and conduct of the study, and also our plans to disseminate the study results to participants and relevant wider patient communities. However, at the end of the study, patients were asked to assess the burden of time required to participate in the study.

Sample size calculation

A sample size was estimated based on the association between sleep quality and DED. With a hospital-based DED prevalence approximating 34%¹⁰ and considering type 1 and 2 errors at 0.05, 0.20, respectively, and a ratio of normal to poor sleep quality of 2:1,¹¹ a total of 1338 patients were required to detect an OR of 1.4.¹²

Data collection

Demographic data (ie, age, sex, educational level and family income) and health risk behaviours (ie, smoking, alcohol consumption, digital screen use and contact lens use) were collected through face-to-face interviews conducted by a single, well-trained interviewer. Data on underlying systemic diseases (ie, hypertension; HT, diabetes mellitus DM, dyslipidaemia; DLP, autoimmune diseases and obstructive sleep apnoea; OSA), ophthalmic diseases (ie, glaucoma and pterygium), a history of eyelid surgery and systemic medication use were retrieved from the hospital electronic medical records. After completing baseline data collection, six standardised questionnaires assessing physical activity, depression, sleep quality and DED were self-administered by the participants, under the supervision of a research assistant. Physical activity was measured using the Global Physical Activity Questionnaire (GPAQ).¹³ The GPAQ is a tool that gathers information on physical activity participation (in three domains) and sedentary behaviour, with a total of 16 questions. Total Metabolic Equivalent of Task (MET) minutes per week were classified as inactive, active and highly active according to the following bands: <600, 600–1200 and >1200 MET-minutes/week, respectively.¹³

Depressive symptoms

Depressive symptoms were assessed using the Patient Health Questionnaire (PHQ-9) that included nine questions, each scored 0–3 with a total score of 27.¹⁴ The total score was categorised into five groups: normal (0–4), mild (5–9), moderate (10–14), moderately severe (15–19) and severe (20–27). Participants were classified with depression for a PHQ-9 score greater than 9.

Sleep factors

Sleep factors, including sleep quality, chronotype and risk of having OSA, were assessed using the Pittsburgh Sleep Quality Index (PSQI),¹⁵ Composite Scale of Morningness (CSM)¹⁶ and STOP-Bang questionnaire,¹⁷ respectively.

The PSQI is a self-rated questionnaire that assesses sleep quality and disturbances over a 1-month interval. The scores range from 0 to 21, and a total score of more than 5 is considered as a significant sleep disturbance.¹⁵ The CSM consists of 13 questions regarding individuals’ preferred wake-up time and bedtime, preferred physical and mental activity times, and subjective alertness. The total score ranges from 13 (ie, extreme eveningness) to 55 (ie, extreme morningness).¹⁶ The STOP-Bang questionnaire comprises eight questions, scored on Yes/No answers. Scores range from 0 to 8, with a score of ≥3 indicating high sensitivity for detecting OSA.¹⁷

Outcome measurements

Dry eye symptoms were evaluated using the Ocular Surface Disease Index (OSDI) questionnaire,¹⁸ which consists of 12 items in 3 subscales (ie, ocular symptoms, vision-related function and environmental triggers). Each item was rated on a 0–4 scale with 0 corresponding to ‘none of the time’ and 4 corresponding to ‘all of the time’. The total score ranges from 0 to 100 with scores between 0 and 12 representing normal, 13 and 22 mild DED, 23 and 32 moderate DED and greater than 33 severe DED.¹⁹

All patients were measured for tear film break-up time (TBUT) using fluorescein strips and evaluated for ocular surface staining using Oxford grading.¹⁹ TBUT is the period from the opening of the eyelid to the first tear breakup and is represented by an average of three times recorded. Tests were conducted bilaterally and the eye with the higher severity was used for DED diagnosis according to the Tear Film and Ocular Surface Society Dry Eye Workshop (DEWS) II criteria, which includes (1) OSDI≥13 and (2) TBUT<10 s or positive surface staining of >5 spots within the cornea.

Statistical analysis

Continuous and categorical data were described using mean and SD or medians with IQR, and frequency with percentages, respectively. Differences between DED and non-DED groups were compared using Student’s t-test or quantile regression for continuous data and the χ² test or Fisher’s exact test for categorical data, as appropriate.

A causal diagram (see figure 1) was constructed, considering depression as the independent variable, sleep quality as the mediator, and DED as the outcome of interest. Mediation analysis was performed using the following steps: First, a mediation model was constructed by regressing sleep quality on depression along with other risk factors. Model selection was performed as follows: a simple logistic regression was performed by fitting each of the factors on sleep quality. Then, factors with p<0.1 were simultaneously included in a multivariable logistic regression. Only significant risk factors were retained in the mediation model. Additionally, four factors previously identified as DED risk factors (age, HT, antihistamine use and digital screen time) were included in the multivariate model. Second, an outcome model was constructed by regressing DED on sleep quality and depression along with other DED risk factors. Model selection was performed using the same process as the mediation model. The direct effect of depression, and its indirect (mediated) effect through sleep quality, on DED were then estimated using the product-of-coefficient method.^{20 21} A bootstrap analysis with 1000 replications was conducted to quantify the mediation effect, which was then averaged across the replications. A 95% CI was also calculated using a bias-corrected bootstrap technique.²² The proportion mediated, which was the ratio of the natural indirect effect and the total effect, was ultimately reported along with 95% CI to demonstrate the relative contribution of sleep quality in the pathway through which depression affects DED. All analyses were performed using Stata V.18.0 (StataCorp).

Figure 1. A constructed causal diagram, considering depression as the independent variable, sleep quality as the mediator and dry eye disease as the outcome of interest.

Results

Of the 1338 patients screened, 17 were excluded due to duplicated records, aged <18 years and missing outcomes, leaving 1321 patients remaining in the analysis. Baseline characteristics are described in table 1. The mean age was 58.3 (SD 14) years with female predominance (73.7%). Over half of patients were educated to degree level or higher (58.7%) with average earnings of 150 000 Thai baht per year (approximately US$4283, 51.4%); 29 patients (2.2%) were current smokers and 172 patients (13.1%) consumed alcohol. Common underlying diseases included DLP in 666 patients (50.4%), DM in 148 patients (11.2%), HT in 81 patients (6.1%), autoimmune diseases in 65 patients (4.9%) and OSA in 52 patients (3.9%). Antihistamines and anticholinergics were used in 161 patients (12.2%) and 149 patients (11.3%), respectively. Glaucoma and pterygium were detected in 295 patients (22.3%) and 67 patients (5.1%), respectively. A small percentage of patients had a history of previous eyelid surgery (2.9%) or contact lens use (2.3%). The median digital screen time per day was 4 hours (IQR 2, 6 hours). Regarding physical activity, 65.5% of patients achieved MET minutes of 600 or more per week and were considered physically active.

Table 1. Baseline patient characteristics (N=1321).

Characteristics	N (%)
Sex
Male/female	347 (26.3)/974 (73.7)
Age, mean (SD)	58.3 (14.0)
BMI, mean (SD)	24.7 (4.4)
Education level
Less than degree level/degree level or higher	540 (41.3)/769 (58.7)
Income (Thai Baht per year)
<150 000/≥150 000	671 (51.4)/635 (48.6)
Current smoker
Yes/No	29 (2.2)/1289 (97.8)
Alcohol consumption
Yes/No	172 (13.1)/1137 (86.9)
Hypertension
Yes/No	81 (6.1)/1240 (93.9)
Diabetes
Yes/No	148 (11.2)/1173 (88.8)
Obstructive sleep apnoea
Yes/No	52 (3.9)/1269 (96.1)
Dyslipidaemia
Yes/No	666 (50.4)/655 (49.6)
Autoimmune disease
Yes/No	65 (4.9)/1254 (95.1%)
Anticholinergics
Yes/No	161 (12.2)/1160 (87.8)
Antihistamine
Yes/No	149 (11.3)/1172 (88.7)
Glaucoma
Yes/No	295 (22.3)/1026 (77.7)
Pterygium
Yes/No	67 (5.1)/1254 (94.9)
History of eyelid surgery
Yes/No	38 (2.9)/1283 (97.1)
Contact lens use
Yes/No	30 (2.3)/1291 (97.7)
Digital screen time (hour), median (IQR)	4.0 (2.0, 6.0)
GPAQ score (MET-minutes per week)
<600/≥600	453 (34.5)/862 (65.5)

BMI, body mass index; GPAQ, Global Physical Activity Questionnaire; MET, Metabolic Equivalent of Task.

Over 40% had a PSQI score >5, indicating poor sleep quality, whereas 107 patients (8.1%) were classified with moderate to severe depression according to the PHQ-9 questionnaire. 64 patients (4.9%) had a STOP-Bang score of 5 or more, indicating a high risk of either moderate or severe OSA. More than half of patients (53.5%) exhibited an intermediate chronotype, indicating no preference for either morning or evening, while 45.8% of patients were morning type, preferring to sleep and wake early and the remaining patients were evening type.

Prevalence of DED

Of 1321 patients, 668 were diagnosed with DED according to the DEWS II criteria, representing a hospital-based prevalence of 50.6% (95% CI: 47.8%, 53.3%).

The median OSDI score was 15 (IQR: 4, 33). Among the 668 patients with DED, 217 (32.5%) had mild DED, 150 (22.5%) had moderate DED and 301 (45.0%) had severe DED. The mean TBUT was 7 s (SD=3.4); 525 patients (39.7%) had a short TBUT of less than 5 s. The median for ocular surface staining was Oxford grade 1 (IQR 0, 1); Oxford grades 2–3 (mild to moderate staining) and grades 4–5 (severe staining) were observed in 214 patients (16.2%) and 13 patients (1%), respectively.

Outcome model: risk factors for DED

Six factors were significantly associated with DED in a univariate analysis, including age, digital screen time, PSQI score, PHQ-9 score, STOP-Bang score and CSM with unadjusted ORs (95% CIs) of 0.99 (0.98, 0.99), 1.05 (1.01, 1.08), 2.75 (2.19, 3.45), 2.58 (1.67, 3.97), 1.81 (1.05, 3.14) and 1.32 (0.11, 1.68), respectively. After adjusting for other covariates (ie, age, HT, antihistamine use and digital screen time), only the PSQI score, PHQ-9 score, STOP-Bang score and CSM remained significantly associated, with adjusted ORs (95% CIs) of 2.48 (1.96, 3.14), 1.65 (1.05, 2.61), 1.81 (1.05, 3.14) and 1.32 (1.04, 1.68), respectively (table 2).

Table 2. Factors associated with DED diagnosis: univariate and multivariate analysis of outcome model.

Characteristics	DED	Non-DED	Unadjusted OR (95% CI)	P value	Adjusted OR*	Adjusted p value
	(n=668)	(n=653)			(95% CI)
Sex
Male	173 (49.86)	174 (50.14)	1	0.757	–	–
Female	495 (50.82)	479 (49.18)	1.04 (0.81, 1.33)
Age, mean (SD)	57.39 (14.28)	59.20 (13.64)	0.99 (0.98, 0.99)	0.018	1.00 (0.99, 1.01)	0.768
BMI, mean (SD)	24.51 (4.44)	24.80 (4.40)	0.98 (0.96, 1.01)	0.231	–	–
Education level
Less than degree level	260 (48.15)	280 (51.85)	1		–	–
Degree level or higher	403 (52.41)	366 (47.59)	1.19 (0.95, 1.48)	0.129
Income (Thai Baht per year)
<150 000	328 (48.88)	343 (51.12)	1		–	–
≥150 000	330 (51.97)	305 (48.03)	1.13 (0.91, 1.41)	0.265
Current smoker
Yes	13 (44.83)	16 (55.17)	1.27 (0.60, 2.66)	0.529	–	–
No	654 (50.74)	635 (49.26)	1
Alcohol consumption
Yes	96 (55.81)	76 (44.19)	1.27 (0.92, 1.76)	0.14	–	–
No	566 (49.78)	571 (50.22)	1
Hypertension
Yes	135 (45.92)	159 (54.08)	0.79 (0.61, 1.02)	0.071	0.77 (0.58, 1.01)	0.059
No	533 (51.90)	494 (48.10)	1		1
Diabetes
Yes	68 (45.95)	80 (54.05)	0.81 (0.58, 1.14)	0.826	–	–
No	600 (51.15)	573 (48.85)	1
Obstructive sleep apnoea
Yes	25 (48.08)	27 (51.92)	0.90 (0.52, 1.57)	0.714	–	–
No	643 (50.67)	626 (49.33)	1
Dyslipidaemia
Yes	337 (50.60)	329 (49.40)	1.00 (0.81, 1.24)	0.981	–	–
No	331 (50.53)	324 (49.47)	1
Autoimmune disease
Yes	35 (53.85)	30 (46.15)	1.15 (0.70, 1.89)	0.588	–	–
No	633 (50.40)	623 (49.60)	1
Anticholinergics
Yes	83 (51.55)	78 (48.45)	1.05 (0.75, 1.45)	0.79	–	–
No	585 (50.43)	575 (49.57)	1
Antihistamine
Yes	86 (57.72)	63 (42.28)	1.38 (0.98, 1.95)	0.065	1.19 (0.83, 1.71)	0.335
No	582 (49.66)	590 (50.34)	1		1
Glaucoma
Yes	158 (53.56)	137 (46.44)	1.12 (0.93, 1.35)	0.242	–	–
No	466 (49.52)	475 (50.48)	1
Pterygium
Yes	77 (52.38)	70 (47.62)	1.08 (0.77, 1.53)	0.648	–	–
No	589 (50.38)	580 (49.62)	1
History of eyelid surgery
Yes	23 (60.53)	15 (39.47)	1.52 (0.78, 2.93)	0.216	–	–
No	645 (50.27)	638 (49.73)
Contact lens use
Yes	17 (56.67)	13 (43.33)	0.78 (0.37, 1.61)	0.5	–	–
No	651 (50.43)	640 (49.57)	1
Digital screen time (hour), median (IQR)	4.0 (2.0, 6.0)	3.0 (1.0, 5.0)	1.05 (1.01, 1.08)	0.007	1.03 (0.99, 1.07)	0.132
GPAQ score (MET-minutes per week)
<600	225 (49.67)	228 (50.33)	0.95 (0.75, 1.19)	0.636	–	–
≥600	440 (51.04)	422 (48.96)	1
PSQI global scores
>5	353 (65.13)	189 (34.87)	2.75 (2.19, 3.45)	<0.001	2.48 (1.96, 3.14)	<0.001
≤5	315 (40.44)	464 (59.56)	1		1
PHQ-9 scores
>9	76 (71.03)	31 (28.97)	2.58 (1.67, 3.97)	<0.001	1.65 (1.05, 2.61)	0.031
≤9	592 (48.76)	622 (51.24)	1		1
STOP-Bang questionnaire
≥5	42 (65.62)	22 (34.38)	1.91 (1.13, 3.24)	0.016	1.81 (1.05, 3.14)	0.033
<5	626 (50.00)	626 (50.00)	1		1
CSM
Evening or intermediate type	403 (56.36)	312 (43.64)	1.67 (1.34, 2.08)	<0.001	1.32 (1.04, 1.68)	0.023
Morning type	264 (43.64)	341 (56.36)	1		1

^*Covariates with a p<0.1 (ie, age, hypertension, antihistamine, digital screen time, PSQI global scores, PHQ-9 scores, STOP-Bang questionnaire and CSM) were simultaneously included in the multivariable logistic regression model.

BMI, body mass index; CSM, Composite Scale of Morningness; DED, dry eye disease; GPAQ, Global Physical Activity Questionnaire; MET, metabolic equivalent of task; PHQ, Patient Health Questionnaire; PSQI, Pittsburgh Sleep Quality Index.

Mediation model: depression and sleep quality

Two equations (depression → sleep quality and sleep quality → DED) were constructed as shown in figure 1 and tables2 3. Depression was associated with a higher likelihood of poor sleep quality (PSQI score >5) compared with patients without depression, with an unadjusted OR of 4.82 (95% CI: 3.07, 7.57). After adjusting for covariates (ie, age, educational level, income level, smoking status, OSA, digital screen time, chronotype and STOP-Bang score), the OR decreased to 3.87, although it remained significant (95% CI: 2.43, 6.15) (see table 3).

Table 3. Factors associated with sleep quality assessed by the Pittsburgh Sleep Quality Index questionnaire: mediation model.

Characteristics	Poor sleep quality(n=542)	Good sleep quality(n=779)	Crude OR(95% CI)	P value	Adjusted OR*(95% CI)	AdjustedP value
Sex
Female	392 (40.2)	582 (59.8)	0.88 (0.69, 1.13)	0.332	–	–
Male	150 (43.2)	197 (56.8)	1
Age, mean (SD)	56.6 (14.4)	59.4 (13.5)	0.99 (0.98, 0.99)	<0.001	1.00 (0.98, 1.05)	0.522
BMI, mean (SD)	24.8 (4.4)	24.6 (4.4)	1.01 (0.99, 1.04)	0.364	1	–
Education level
Less than degree level	204 (37.8)	336 (62.2)	0.79 (0.63, 0.98)	0.036	0.94 (0.72, 1.22)	0.634
Degree level or higher	335 (43.6)	434 (56.4)	1		1
Income (Thai Baht per year)
<150 000	255 (38.0)	416 (62.0)	0.77 (0.62, 0.96)	0.022	0.83 (0.64, 1.07)	0.149
≥150 000	281 (44.3)	354 (55.7)	1		1
Current smoker
Yes	18 (62.1)	11 (37.9)	2.40 (1.12, 5.13)	0.023	1.83 (0.81, 4.15)	0.145
No	522 (40.5)	767 (59.5)	1		1
Alcohol consumption
Yes	77 (44.8)	95 (55.2)	1.19 (0.86, 1.65)	0.285	–	–
No	460 (40.5)	677 (59.5)	1
Hypertension
Yes	124 (42.2)	170 (57.8)	1.06 (0.82, 1.38)	0.650	–	–
No	418 (40.7)	609 (59.3)	1
Diabetes
Yes	56 (37.8)	92 (62.2)	0.86 (0.61, 1.22)	0.403	–	–
No	486 (41.4)	687 (58.6)	1
Chronic kidney disease
Yes	55 (40.7)	80 (59.3)	0.97 (0.69, 1.42)	0.943	–	–
No	487 (41.1)	699 (58.9)	1
Obstructive sleep apnoea
Yes	29 (55.8)	23 (44.2)	1.86 (1.06, 3.25)	0.027	1.76 (0.97, 3.21)	0.063
No	513 (40.4)	756 (59.6)	1		1
Dyslipidaemia
Yes	276 (41.4)	390 (58.6)	1.03 (0.83, 1.29)	0.759	–	–
No	266 (40.6)	389 (59.4)	1
Autoimmune disease
Yes	33 (50.8)	32 (49.2)	1.51 (0.92, 2.49)	0.102	–	–
No	509 (40.5)	747 (59.5)	1
Digital screen time (hour), median (IQR)	4.0 (2.0, 6.0)	3.0 (1.0, 5.0)	1.05 (1.02, 1.09)	0.002	1.01 (0.97, 1.05)	0.552
GPAQ score (MET-minutes per week)
<600	178 (39.3)	275 (60.7)	0.90 (0.71, 1.13)	0.365	–	–
≥600	361 (41.9)	501 (58.1)	1
PHQ-9 scores
>9	80 (74.8)	27 (25.2)	4.82 (3.07, 7.57)	<0.001	3.87 (2.43, 6.15)	<0.001
≤9	462 (38.1)	752 (61.9)	1		1
STOP-Bang questionnaire
≥5	33 (51.6)	31 (48.4)	1.57 (0.95, 2.59)	0.077	1.40 (0.82, 2.40)	0.212
<5	506 (40.4)	746 (59.6)	1		1
CSM
Evening or intermediate type	357 (49.9)	358 (50.1)	2.26 (1.80, 2.84)	<0.001	1.88 (1.48, 2.42)	<0.001
Morning type	185 (30.6)	420 (69.4)	1		1

^*Covariates with a p<0.1 (ie, age, education levels, income, current smoker, obstructive sleep apnoea, digital screen time, PHQ-9 scores, STOP-Bang questionnaire and CSM) were simultaneously included in the multivariable logistic regression model.

BMI, body mass index; CSM, Composite Scale of Morningness; GPAQ, Global Physical Activity Questionnaire; MET, Metabolic Equivalent of Task; PHQ, Patient Health Questionnaire.

Mediation analysis that included a 1000 bootstrap replication indicated patients with depression were 2.20 (95% CI: 1.41, 3.41) times more likely to have DED compared with patients without depression (see table 4). The total effect was decomposed into an indirect effect (ie, mediation) through poor sleep quality (OR=1.32; 95% CI: 1.18, 1.49) and a direct effect (OR=1.66; 95% CI: 1.06, 2.58). The indirect effect of depression via poor sleep quality was estimated to account for 34% (95% CI: 9.4%, 58.6%) of DED, with the remaining 66% due to other mechanisms.

Table 4. Estimation of direct and indirect effects of depression on dry eye disease through poor sleep quality: a mediation analysis.

Pathway	Coef.	95% CI	P value	OR	95% CI
Direct effect:Depression → DED	0.12	0.02, 0.23	0.021	1.66	1.06, 2.58
Indirect effect:Depression → poor sleep quality → DED	0.06	0.04, 0.09	<0.001	1.32	1.18, 1.49
Total effect	0.19	0.04, 0.09	<0.001	2.20	1.41, 3.41

DED, dry eye disease.

Discussion

We estimated a DED prevalence of 50.6% in our cross-sectional hospital outpatient study. Depression and poor sleep quality were significantly associated with DED, in addition to age, duration of digital screen use and chronotype. The result of the mediation analysis indicated that depression increased DED risk, partly through direct and indirect effects, with the latter mediated through the pathway of poor sleep quality.

We found that depression (OR=2.58, 95% CI: 1.67, 3.97) and sleep quality (OR=2.75, 95% CI: 2.19, 3.45) showed a strong association with DED (OR>2.0). A previous meta-analysis summarising results from 10 studies published between 1989 and 2015 demonstrated a similar magnitude of association between depression and DED with an OR of 2.92 (95% CI: 2.13, 4.01).²³

Regarding sleep quality, patients with DED in our study tended to have higher PSQI scores compared with patients without DED (mean difference; MD=1.4, 95% CI: 1.6, 1.1). This finding aligns with a previous meta-analysis of six studies, which reported a weighted MD of 1.69 (95% CI: 0.82, 2.56) in PSQI scores.¹² A community-based study in China including 3070 participants aged 18–80 years also demonstrated a strong association between poor sleep quality and increased severity of DED.²⁴ They found that worse PSQI scores were significantly associated with poorer OSDI scores. Additionally, short sleep duration was also associated with increased DED prevalence in a dose-response fashion, with a 1.20-fold increase in patients with sleep duration of <5 hours per day and a 1.29-fold increase in patients with sleep duration of ≤4 hours per day.²⁵

Previous evidence has demonstrated that short sleep duration, psychological stress and a history of depressed mood were significantly associated with dry eye symptoms, even after considering patients’ characteristics, lifestyle and other medical factors.²⁶ However, the mechanism underlying the complex association between depression, poor sleep quality and DED remains uncertain. Based on our data, approximately 70% of patients with depression (PHQ-9 score >9) had DED, compared with only 48% of non-depressed individuals. To investigate this, we performed a mediation analysis to test the hypothesis that depression increases the risk of DED at least partially via poor sleep quality. Our findings suggest a significant causal link between depression and DED, with 34% of the indirect effect mediated by poor sleep quality and the remaining 66% attributed to other mechanisms.

A mouse study revealed that sleep deprivation could substantially impair the functionality of the lacrimal glands by approximately 50% of normal levels after 2 days of deprivation, with this reduction persisting for 10 days.²⁷ Tang et al further demonstrated that sleep deprivation induced DED at a cellular level through abnormal superficial epithelial corneal cell morphology caused by downregulation of PPARα and TRPV6 expression.²⁸ A clinical study of 20 healthy male human subjects showed that 24-hour sleep deprivation could induce DED characterised by significantly increased tear film hyperosmolarity, reduced tear secretion and shortened TBUT.⁹ A recent cohort study has also demonstrated that the 8-year incidence of DED was significantly higher in participants with sleep disorders, with a threefold increased risk of DED associated with sleep disorders.²⁹ This evidence is strongly indicative of a causal relationship between poor sleep quality and DED.

The additional pathways underlying the causal relationship between depression and DED remain largely unexplored. Nevertheless, inflammatory cytokines, notably IL-6 and TNF-α, have been shown to be pivotal in the pathogenesis of depression and are markedly elevated in the tears of patients with comorbid depression and DED compared with healthy controls, indicative of a significant inflammatory component in the interplay between both conditions.³⁰

Meanwhile, the relationship between DED and depression or sleep disorders may be bidirectional, meaning depression cannot only serve as a potential cause, but also a consequence of DED, and vice versa. An umbrella review on the association between DED and depression suggested that the ocular discomfort and impact on daily life caused by DED can negatively affect mood and overall well-being, possibly leading to or exacerbating depression.³¹ Additionally, another study reported that patients with DED were more likely to experience anxiety and depression compared with healthy individuals.³² Similarly, a recent meta-analysis showed that individuals with DED have poorer sleep quality than the general population, characterised by lower subjective sleep quality, longer sleep latency, and a higher risk of abnormal sleep durations such as insufficient or excessive sleep.³³ These findings indicate that DED may contribute to the development of sleep disorders as well. Further studies are needed to clarify the nature of these associations.

Increased DED prevalence has been shown in association with digital screen use, including visual display terminal (VDT) users and office workers, ranging from 26% to 70%.³⁴ Interestingly, even short periods of VDT use of as little as 1–2 hours per day can be associated with increased DED risk. Our study’s median digital screen time was 3 hours daily (IQR 2, 6 hours per day) and we observed a significant negative correlation between age and digital screen time, which may explain why younger patients were more likely to have DED compared with older patients. However, both age and duration of digital screen use became non-significant following adjustment for chronotype, sleep quality and severity of depression.

Chronotype refers to an individual’s circadian typology or physiologic preference to sleep and alertness at specific times of the day. There are three main types: morning, intermediate and evening types. The evening type is associated with sleep disturbances, poor lifestyle habits and an increased risk of psychiatric disorders and cardiovascular diseases.^{35 36} Based on our study subjects, the most common chronotype was intermediate (53.5%), followed by morning (45.8%) and finally evening type (0.7%). Patients with intermediate and evening types tended to have a higher DED risk compared with morning type (OR=1.67), supporting previous reports.³⁷ It has been hypothesised that the association between chronotype and DED could be related to neurotransmitters, especially 5-hydroxytryptamine (5-HT, serotonin), which is known to be involved in the sensitisation of nociceptors and is present in human tears.³⁸ This hypothesis is further supported by the higher level of 5-HT in tears detected in symptomatic DED individuals compared with unaffected controls.³⁹

This study has some limitations. First, it was conducted in an outpatient eye clinic of a tertiary hospital centre, limiting the generalisability of the findings to a broader population. Second, a causal relationship could not be definitively established due to the nature of the cross-sectional study design. Lastly, this study did not incorporate objective assessments of sleep quality, such as actigraphy, melatonin measurement or polysomnography. Nevertheless, PSQI has been widely used in sleep quality research and has demonstrated a diagnostic sensitivity of 89.6% and specificity of 86.5% in distinguishing between good and poor sleepers when compared with polysomnography, which is considered the reference standard for sleep assessment.¹⁵

In summary, DED was detected in approximately half of the Thai patients attending the ophthalmology outpatient clinic. Depression and poor sleep quality are significant risk factors for DED. Notably, poor sleep quality not only directly exacerbates DED but also mediates the relationship between depression and DED. Therefore, assessment of sleep quality and depression status should be adopted in routine clinical practice in the management of DED patients. Larger-scale cohort studies using objective sleep measurements and tear cytokine analysis are imperative to elaborate on the causal links between depression, sleep quality and DED.

This funder had no role in the design of this study and will not have any role during its execution, analyses, interpretation of the data or decision to submit results.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.