Apparent protective associations between smoking, alcohol consumption, and dry eye disease: evidence from a community-based Adult Dry Eye Cohort (ADEC) in Beijing

Abstract

Background

To investigate the associations of current smoking and current alcohol consumption with dry eye disease (DED) in a community-based population of middle-aged and older adults, and to determine whether these associations persisted after accounting for demographic, lifestyle, environmental, and psychological factors.

Methods

This cross-sectional analysis was conducted within the Beijing Adult Dry Eye Cohort (ADEC) and included 813 participants aged 45 years and older with complete baseline clinical and questionnaire data. DED was defined as an Ocular Surface Disease Index (OSDI) score of 13 or higher together with a non-invasive tear break-up time (NIBUT) of less than 10 s. NIBUT, tear meniscus height, tear meniscus area, and meibomian gland loss were evaluated as objective ocular-surface parameters. Demographic characteristics, lifestyle factors, and environmental exposures were collected using structured questionnaires, while anxiety and depression were assessed using validated scales. Associations of current smoking and current alcohol consumption with DED were examined using logistic regression models before and after multivariable adjustment.

Results

Among 813 participants, 344 were classified as having DED, corresponding to a prevalence of 42.3%. In the unadjusted analysis, current smoking was associated with lower odds of DED; however, this association was no longer significant after adjustment for demographic, environmental, lifestyle-related, and psychological factors. Current alcohol consumption was not independently associated with DED. Differences in objective ocular-surface parameters were observed across smoking and alcohol-consumption groups. Participants who smoked or consumed alcohol showed longer NIBUT, but also greater meibomian gland loss, indicating that functional and structural ocular-surface parameters did not vary in a consistent direction.

Conclusions

Current smoking and alcohol consumption should not be interpreted as protective against DED in clinical or public health contexts. Isolated favorable tear-film findings should be evaluated together with meibomian gland status and other ocular-surface features. These findings support comprehensive DED risk assessment and reinforce the importance of smoking cessation and avoidance of excessive alcohol consumption in public health messaging.

Introduction

According to the TFOS DEWS III report, dry eye is a multifactorial, symptomatic disease characterized by a loss of homeostasis of the tear film and/or ocular surface, in which tear film instability and hyperosmolarity, ocular surface inflammation and damage, and neurosensory abnormalities are etiological factors [1]. Reported prevalence of dry eye disease varies substantially according to diagnostic criteria, age, sex, and the population studied. Current evidence summarized in TFOS DEWS III further indicates that DED risk factors span demographic, environmental, systemic, iatrogenic, and lifestyle-related domains [2]. DED not only impairs quality of life but also imposes a substantial socioeconomic burden in China and globally [3, 4]. A recent systematic review and meta-analysis further reported a high prevalence and incidence of dry eye disease across Asian populations, underscoring its growing public health relevance in this region [5].

Among modifiable lifestyle behaviors, smoking and alcohol consumption are of particular interest because of their widespread prevalence and potential biological impact on ocular health. Tobacco smoke contains numerous oxidants and pro-inflammatory agents that can disrupt tear film stability and damage the lacrimal and meibomian glands [6–12]. However, some population studies have reported no association or even inverse associations between smoking and DED [13, 14], raising the possibility that these findings may be influenced by confounding from demographic, occupational, behavioral, and psychological factors, rather than reflecting a true independent association. The role of alcohol is similarly complex: moderate intake has been linked to antioxidant and anti-inflammatory actions [15], whereas chronic heavy consumption may impair tear secretion and increase oxidative stress [16, 17]. Clarifying these relationships is essential for accurately evaluating the ocular effects of smoking and alcohol consumption and for informing prevention strategies.

Recent TFOS DEWS III evidence emphasizes that age is an important determinant of DED epidemiology, with DED and meibomian gland dysfunction becoming increasingly relevant after middle age [2]. In addition, population-based evidence among older adults has shown that both signs and symptoms of DED are associated with poorer health-related quality of life [18]. These findings support focusing on middle-aged and older adults as a population of particular public health relevance.

Using baseline data from the Beijing Adult Dry Eye Cohort (ADEC), we investigated the associations of smoking and alcohol consumption with DED among participants aged 45 years and older. Particular emphasis was placed on whether any apparent inverse associations persisted after comprehensive adjustment for demographic, environmental, behavioral, and psychological confounders, and whether these exposures showed consistent patterns across objective ocular-surface parameters.

Materials and methods

Study design and setting

The ADEC study is an ongoing population-based cohort study. Baseline recruitment was conducted between July 15 and August 31, 2023. The present analysis is based on cross-sectional data collected at baseline. Participants were selected using a cluster random sampling method. Briefly, three streets encompassing 15 residential communities in Beijing, China, were included as sampling units. A total of 1,143 individuals were registered during the baseline community survey. One individual completed only the registration step and did not undergo baseline ocular examinations or questionnaire assessments. Although age and sex were available from the registration record, DED status and key exposure and covariate data were unavailable. This record was therefore excluded from the analytic dataset. The remaining 1,142 participants constituted the cleaned baseline dataset. The present analysis was then restricted to participants aged 45 years or older, resulting in an analytic cohort of 813 participants. In this analysis, the threshold of 45 years was used as the operational definition of middle-aged and older adults, consistent with its use in population-based aging research in China [19].

A one-year follow-up was subsequently completed between July 17 and August 22, 2024; however, follow-up data were not included in the present analysis.

Human ethics and consent to participate declarations

This observational cohort study was approved by the Ethics Committee of Beijing Tongren Hospital, Capital Medical University (Ethics approval number: TREC2024-KY084), and was conducted in accordance with the principles of the Declaration of Helsinki. Signed written informed consent was obtained from each participant after explanation of the nature and possible consequences of the study. This study was registered at ClinicalTrials.gov (https://clinicaltrials.gov/). The Clinical Trial Number is NCT06621446, and the Registration Date was September 23, 2024.

Inclusion and exclusion criteria

Inclusion Criteria: 1) Aged 45 years or older; 2) resident population of the sample community; 3) Able to complete dry eye examination; 4) Informed consent signed.

Exclusion Criteria: 1) Mental illness: clinically significant psychiatric or cognitive conditions that impaired the participant’s ability to understand the study procedures, provide reliable questionnaire responses, or cooperate with ocular examinations; 2) Mobility difficulties: difficulties in transferring, walking, moving, or maintaining the seated examination posture required for ocular testing; 3) Other conditions that would preclude completion of baseline dry eye examinations or questionnaires (to be determined by the study physician).

These operational definitions were framed according to their impact on study participation and examination completion, with reference to the WHO definition of mental disorders [20] and the WHO International Classification of Functioning, Disability and Health framework [21] for mobility-related functioning.

Exclusion criteria were assessed by the study physician based on direct observation, communication with the participant, and, when necessary, information provided by accompanying family members or community staff. These criteria were used operationally to determine whether participants could provide reliable questionnaire responses and complete ocular examinations, rather than to establish formal diagnoses of mental illness or mobility disorders.

Outcomes and measures

All ocular examinations were performed on-site at each participating community center using a standardized mobile examination protocol. To minimize diurnal variation, examinations were uniformly scheduled between 8:30 AM and 11:30 AM. Room temperature and relative humidity were monitored in real time and maintained at 22 ± 2 ℃ and 45 ± 5%, respectively. Participants rested in the room for 10 min prior to examination. Participants were asked to refrain from using any topical ophthalmic medications (including artificial tears) for at least 2 h before testing. Participants who reported contact lens wear were required to discontinue lens wear for at least 24 h before dry eye testing. The order of testing was standardized: tear meniscus height, tear meniscus area, and non-invasive tear break-up time were performed first, followed by corneal fluorescein staining and meibomian gland imaging. All instruments were calibrated daily, and measurements were performed by trained technicians. This standardized protocol was consistently followed at both baseline and the 1-year follow-up.

A dry eye and corneal topography tester (MediWorks, Shanghai, China, www.mediworks.com.cn, DEA520) was used to perform a comprehensive dry eye examination, including non-invasive tear break-up time (NIBUT), tear meniscus height, morphological evaluation of meibomian gland, corneal fluorescein sodium staining, and the built-in OSDI questionnaire. NIBUT and tear meniscus height were automatically measured using the device software. The OSDI questionnaire was not used to screen participants for eligibility; it was administered as part of the baseline assessment and was used, together with NIBUT, to define DED status in the present analysis.

The diagnosis of dry eye disease was based on the presence of both symptoms and tear-film instability, consistent with the TFOS DEWS II diagnostic framework [22]. Participants were classified as having DED if they had a positive symptom score, defined as an OSDI score of 13 or higher, together with an objective sign of tear-film instability, defined as a non-invasive tear break-up time of less than 10 s. OSDI scores were additionally used to describe symptom severity as normal, mild, moderate, or severe, corresponding to scores of 0–12, 13–22, 23–32, and 33 or higher, respectively. These OSDI-based severity categories were used only for descriptive presentation and were not used as the outcome variable in the regression analyses.

Assessment of smoking and alcohol consumption

Questionnaire data were collected face-to-face by trained investigators and entered through the same electronic questionnaire module used during the dry eye assessment. The module included an Ocular Surface Disease Index (OSDI) section for dry eye symptom evaluation, together with structured risk-factor items covering smoking status, alcohol consumption, and other demographic, environmental, lifestyle, and psychological factors. For the present analysis, smoking status was assessed using the following question: “Do you currently smoke, including smoking within the past month?” The response options were: (1) No; (2) Yes. Alcohol consumption was assessed using the following question: “Do you currently drink alcohol, including alcohol consumption within the past month?” The response options were: (1) No; (2) Yes. All responses were self-reported. Participants who answered “Yes” were classified as current smokers or current drinkers, whereas those who answered “No” were classified as non-current smokers or non-current drinkers.

Assessment of possible confounding factors

Baseline questionnaires collected demographic and lifestyle information, including: (1) Demographics and medical history: age, sex, hypertension, diabetes. (2) Work environment: long-term exposure to dust, heat, or outdoor conditions over the past five years. (3) Lifestyle factors: sleep quality (self-rated five-point scale), frequency of electronic-screen use, dietary habits (fruit and vegetable intake), and use of ocular lubricants. Sleep quality was collected as one of the structured risk-factor items within the same electronic questionnaire module used during the dry eye assessment. It was assessed using a single self-rated five-category item: very good sleep quality, occasional insomnia or staying up late, frequent insomnia or staying up late, poor sleep quality most of the time, and very poor sleep quality every day. This item was not derived from a validated sleep-quality questionnaire and was analyzed as an ordinal variable in the regression models. (4) Psychological status: anxiety and depression were assessed using the validated Hamilton Anxiety Scale (HAM-A) and Hamilton Depression Scale (HAM-D). For descriptive comparisons, anxiety symptoms were categorized as present when the HAM-A score was 7 or higher, and depressive symptoms were categorized as present when the HAM-D score was 8 or higher. In the logistic regression models, HAM-A and HAM-D scores were entered as continuous variables.

Statistical analysis

All statistical analyses were conducted using SPSS version 27.0 (IBM Corp., Armonk, NY), and a p-value < 0.05 was considered statistically significant. The analysis was structured as follows. The Shapiro–Wilk test was used to assess the normality of continuous variables. Continuous variables were expressed as mean ± standard deviation (SD) if normally distributed, or as median with interquartile range (IQR) if non-normally distributed. Categorical variables were presented as frequencies and percentages.

For binary logistic regression, the outcome variable was DED status as a dichotomous variable. Participants were classified as having DED if they had an OSDI score of 13 or higher together with a non-invasive tear break-up time of less than 10 s. The four OSDI categories were used only to describe symptom severity and were not used as the outcome variable in the regression analyses.

The chi-square test was used to compare categorical variables between groups. Current smoking and current alcohol consumption were included as primary exposure variables. Covariates with statistically significant differences in univariate comparisons were included in the multivariable logistic regression model.

The effects of smoking and alcohol consumption on continuous objective parameters of DED, such as NIBUT, tear meniscus height (TMH), tear meniscus area (TMA), and the proportion of meibomian gland loss (upper and lower eyelids), were analyzed using the Mann-Whitney U test because these variables did not satisfy the assumptions for parametric tests.

Results

Of the 1,142 participants included in the cleaned baseline dataset, the prevalence of DED in the overall baseline population was 35.2%. The mean age of the overall baseline population was 55.70 ± 17.93 years. The analyzed cohort included 813 participants aged 45 years or older, with a mean age of 64.84 ± 9.47 years (range: 45–90 years). Among them, 344 participants were classified as having DED, corresponding to a prevalence of 42.3%. Based on OSDI scores, the proportions of participants with mild, moderate, and severe dry eye symptoms were 27.6%, 10.7%, and 5.2%, respectively. The baseline characteristics of the analyzed cohort are presented in Table 1.

Table 1.

Characteristics of the analyzed cohort aged 45–90 years

	n	%
Gender
-Male	268	33.0%
-Female	545	67.0%
Smoking
-Current	98	12.1%
-Non-current	715	87.9%
Alcohol Consumption
-Current	132	16.2%
-Non-current	681	83.8%
Dry Eye Disease
No	469	57.7%
Yes	344	42.3%
OSDI-based Symptom Severity
Normal	460	56.6%
Mild	224	27.6%
Moderate	87	10.7%
Severe	42	5.2%

Univariate analysis showed that smoking, gender, heated and dusty environments, sleep quality, electronic-screen use, vegetable and fruit intake, anxiety, and depression were significantly associated with DED (Table 2).

Table 2.

Comparison of baseline characteristics between DED and non-DED groups (n = 813)

Variables	n	Non-DED n (%)	DED n (%)	χ²	P value
Smoking				12.888	< 0.001
-Current	98	73(15.6)	25(7.3)
-Non-current	715	396(84.4)	319(92.7)
Alcohol Consumption				3.597	0.058
-Current	132	86(18.3)	46(13.4)
-Non-current	681	383(81.7)	298(86.6)
Gender				15.890	< 0.001
-Male	268	181(38.6)	87(25.3)
-Female	545	288(61.4)	257(74.7)
Indoor environment				0.001	0.976
-Yes	354	204(43.5)	150(43.6)
-No	459	265(56.5)	194(56.4)
Outdoor environment				2.109	0.146
-Yes	33	15(3.2)	18(5.2)
-No	780	454(96.8)	326(94.8)
Heated environment				5.468	0.019
-Yes	132	64(13.6)	68(19.8)
-No	681	405(86.4)	276(80.2)
Dusty environment				6.086	0.014
-Yes	55	23(4.9)	32(9.3)
-No	758	446(95.1)	312(90.7)
Drug use				0.060	0.806
-Yes	342	199(42.4)	143(41.6)
-No	471	270(57.6)	201(58.4)
Sleep Quality				21.815	< 0.001
-Excellent	364	238(50.7)	126(36.6)
-Fairly good	226	128(27.3)	98(28.5)
-Moderate	94	46(9.8)	48(14.0)
-Poor	98	42(9.0)	56(16.3)
-Extremely poor	31	15(3.2)	16(4.7)
Frequency of using electronic screens				11.558	0.003
-Occasional use	131	92(19.6)	39(11.3)
-Frequent use	74	36(7.7)	38(11.0)
-Prolonged daily use	608	341(72.7)	267(77.6)
Vitamin A				0.569	0.752
-Nearly every day	43	26(5.5)	17(4.9)
-Fairly frequent	35	22(4.7)	13(3.8)
-Occasionally	735	421(89.8)	314(91.3)
Vegetables				61.359	< 0.001
-Nearly every day	654	409(87.2)	245(71.2)
-Fairly frequent	94	52(11.1)	42(12.2)
-Occasionally	65	8(1.7)	57(16.6)
Fruits				24.458	< 0.001
-Nearly every day	608	364(77.6)	244(70.9)
-Fairly frequent	116	75(16.0)	41(11.9)
-Occasionally	89	30(6.4)	59(17.2)
Anxiety				397.000	< 0.001
-Yes	239	10(2.1)	229(66.6)
-No	574	459(97.9)	115(33.4)
Depression				164.139	< 0.001
-Yes	470	182(38.8)	288(83.7)
-No	343	287(61.2)	56(16.3)

Further analysis of ocular parameters is shown in Table 3. Smokers had longer NIBUT (first and average) and larger TMH and tear meniscus area than non-smokers (P < 0.05). However, they also had a higher rate of meibomian gland loss in the lower eyelid (P = 0.002). Drinkers had longer average NIBUT but greater lower eyelid meibomian gland loss compared with non-drinkers (P < 0.05).

Table 3.

Association between smoking, alcohol consumption, and ocular-surface parameters

	NIBUT -first	NIBUT -average	TMH (mm)	TMA (mm2)	Meibomian gland loss-upper	Meibomian gland loss-lower
Smoking
-Current	10.5 (8.2,11.7)	12.1 (9.4,14.18)	0.18 (0.15,0.24)	1.81 (1.32,2.56)	0.393 (0.301,0.511)	0.538 (0.371,0.716)
-Non-current	10.0 (5.9,11.7)	11.1 (7.6,13.6)	0.16 (0.14,0.20)	1.63 (1.30,2.18)	0.377 (0.280,0.503)	0.441 (0.289,0.645)
Z	-2.317	-2.478	-3.157	-2.067	-1.036	-3.046
P	0.021	0.013	0.002	0.039	0.300	0.002
Alcohol Consumption
-Current	10.4 (6.4,11.8)	11.9 (9.1,13.9)	0.17 (0.15,0.21)	1.75 (1.30,2.34)	0.411 (0.291,0.514)	0.488 (0.365,0.710)
-Non-current	10.0 (6.0,11.6)	11.1 (7.6,13.7)	0.16 (0.14,0.20)	1.64 (1.31,2.19)	0.375 (0.280,0.500)	0.447 (0.289,0.643)
Z	-1.478	-2.058	-1.938	-1.080	-1.931	-2.047
P	0.139	0.040	0.053	0.280	0.054	0.041

Data are presented as median (IQR)

The results of logistic regression analyses are shown in Table 4. In the univariate analysis, smoking was associated with lower odds of DED (OR = 0.43, 95% CI: 0.26–0.69; P < 0.001), whereas alcohol consumption was not significantly associated with DED (OR = 0.69, 95% CI: 0.47–1.01; P = 0.059). After adjustment for potential confounders, neither smoking nor alcohol consumption remained significantly associated with DED (smoking: OR = 0.83, 95% CI: 0.38–1.81; P = 0.634; alcohol consumption: OR = 1.22, 95% CI: 0.62–2.41; P = 0.560). In the multivariable model, older age, female sex, heated environment, anxiety, and depression were independently associated with higher odds of DED, whereas higher vegetable intake was associated with lower odds of DED.

Table 4.

Binary logistic regression analysis of factors associated with DED

Factors	Univariate logistic regression		Multivariable logistic regression
	OR (95% CI)	P	OR (95% CI)	P
Smoking	0.43(0.26,0.69)	< 0.001	0.83(0.38,1.81)	0.634
Alcohol Consumption	0.69(0.47,1.01)	0.059	1.22(0.62,2.41)	0.560
Age	1.02(1.00,1.03)	0.024	1.04(1.01,1.06)	0.006
Gender	1.86(1.37,2.52)	< 0.001	1.92(1.07,3.45)	0.029
Heated environment	1.56(1.07,2.27)	0.020	2.31(1.23,4.33)	0.009
Dusty environment	1.99(1.14,3.46)	0.015	1.30(0.55,3.05)	0.548
Sleep Quality	1.31(1.16,1.47)	< 0.001	1.11(0.92,1.34)	0.274
Frequency of using electronic screens	1.27(1.05,1.54)	0.014	1.30(0.96,1.78)	0.095
Vegetables	0.41(0.31,0.53)	< 0.001	0.49(0.29,0.82)	0.007
Fruits	0.68(0.56,0.84)	< 0.001	0.99(0.64,1.53)	0.959
Anxiety	2.15(1.93,2.40)	< 0.001	2.01(1.78,2.27)	< 0.001
Depression	1.66(1.54,1.80)	< 0.001	1.65(1.49,1.84)	< 0.001

Reference categories for categorical variables were non-current smoking, non-current alcohol consumption, male sex, no heated environment, and no dusty environment. Sleep quality, frequency of electronic screen use, vegetable intake, and fruit intake were entered as ordinal variables, with odds ratios representing the change in odds associated with a one-category increase. Anxiety and depression scores were entered as continuous variables, with odds ratios representing the change in odds associated with a one-point increase in the corresponding score

Discussion

Interpretation of the main findings

This community-based study examined the associations between smoking, alcohol consumption, and DED in an adult population. Although apparent protective associations were observed in unadjusted analyses, these associations were no longer evident after comprehensive adjustment for confounding factors. This finding indicates that previously reported inverse relationships between smoking, alcohol consumption, and DED may primarily reflect differences in demographic characteristics, lifestyle behaviors, and psychological status rather than true protective effects [13, 23].

Observational analyses of common behavioral exposures are inherently vulnerable to residual confounding, selection bias, exposure misclassification, and heterogeneity in participant characteristics and covariate adjustment [24]. In the present study, the inverse association between current smoking and DED observed in the unadjusted analysis was attenuated after multivariable adjustment, suggesting that the crude association was likely explained, at least in part, by differences in demographic, behavioral, environmental, and psychological characteristics between smokers and non-smokers, rather than reflecting a true independent protective effect. These findings underscore the importance of rigorous confounder control and cautious causal interpretation when evaluating counterintuitive associations in observational studies.

In the unadjusted analysis, current smoking was associated with a lower prevalence of DED, whereas current alcohol consumption showed no statistically significant association. However, individuals who reported smoking or alcohol consumption differed systematically from non-users in several key characteristics, including sex distribution, sleep quality, screen exposure, and psychological status—factors that have been consistently associated with DED in previous studies [25, 26]. After adjustment for these variables, the inverse associations were no longer observed, indicating that smoking and alcohol consumption may function as markers of broader lifestyle patterns rather than independent protective factors. Similar heterogeneity has also been reported in earlier population-based studies, including the Beaver Dam Offspring Study and Lifelines cohort analyses. The Beaver Dam Offspring Study found no significant association between dry eye and smoking history or alcohol consumption, whereas Lifelines analyses reported apparently inverse or sex-specific associations for current smoking and alcohol intake. These findings suggest that such associations should be interpreted cautiously in light of differences in population characteristics, exposure definitions, outcome definitions, and covariate structure [13, 23, 27].

Sleep quality also showed a significant association with DED in the univariate analysis but was not independently associated with DED after multivariable adjustment. Previous evidence has suggested that poorer sleep quality and sleep-related disturbances are associated with higher dry eye symptom scores or an increased risk of DED [28]. In the present study, the attenuation of this association after adjustment suggests that the crude association may have been partly explained by other demographic, behavioral, and psychological covariates included in the multivariable model. Given that sleep quality was assessed using a single self-rated item, this finding should be interpreted cautiously.

Influence of psychological factors

Psychological factors appeared to play an important role in shaping the observed associations in this study. Anxiety and depression were independently associated with DED after multivariable adjustment, consistent with previous reports demonstrating close links between psychological status and DED [29]. Studies evaluating dry eye symptoms and signs concurrently have further shown that psychological distress is more strongly related to subjective symptom burden than to objective ocular-surface findings [30]. This overall pattern is also supported by a systematic review and meta-analysis of psychiatric comorbidity in DED [31]. These observations suggest that psychological factors may influence symptom perception and reporting, and should therefore be taken into account when interpreting lifestyle-related associations with DED.

Psychological distress has been shown to affect pain sensitivity, symptom awareness, and health-seeking behavior, potentially leading to differential reporting of ocular discomfort [32–34]. Accordingly, individuals with lower levels of psychological distress may be less likely to perceive or report dry eye symptoms, independent of objective ocular surface changes. Failure to adequately account for these factors may therefore contribute to misleading conclusions regarding lifestyle exposures and DED risk in epidemiological research.

Interpretation of objective findings in observational studies

Secondary analyses suggested differences in objective ocular surface parameters between exposure groups. Smokers and drinkers demonstrated longer tear film stability measures but also exhibited greater meibomian gland loss. This pattern should be interpreted in light of the multifactorial nature of NIBUT variation. A related analysis of the same ADEC baseline dataset showed that NIBUT was independently associated with demographic, ocular, systemic, and psychological factors after multivariable adjustment [35]. This discordant pattern should be interpreted cautiously, as previous clinical and experimental studies have reported adverse effects of cigarette smoke and alcohol exposure on the tear-film lipid layer, meibomian glands, and ocular surface [16, 17, 36]. These results suggest that short-term functional changes may coexist with longer-term structural alterations. From a population health perspective, however, such findings do not support a protective role of smoking or alcohol consumption. Instead, they underscore the complexity of interpreting subclinical physiological parameters in observational studies and caution against equating transient functional measures with long-term ocular health.

Comparison with previous population-based studies

First, the Beijing Adult Dry Eye Cohort, as a community-based cohort with follow-up assessment, provides a population-based research platform for future evaluation of DED-related risk factors and dynamic ocular-surface changes over time. At the current stage, the core value of the present analysis is to establish a well-characterized baseline reference framework for this cohort.

Second, compared with many previous large population-based studies that mainly relied on symptom questionnaires or self-reported diagnosis, the strength of the present study lies in its integration of systematic objective ocular-surface parameters and multidimensional questionnaire-based variables. DED was defined by the presence of both dry eye symptoms and tear-film instability. We also evaluated objective indicators including NIBUT, tear meniscus height and area, and meibomian gland loss, and incorporated covariates including demographic characteristics, environmental exposure, sleep quality, screen use, dietary patterns, and psychological status. This multidimensional design allowed us not only to assess the associations of smoking and alcohol consumption with DED, but also to further examine whether apparently favorable associations corresponded to a consistent objective ocular-surface phenotype.

Previous population-based studies have reported heterogeneous findings regarding the associations of smoking and alcohol consumption with DED, including increased risk in certain subgroups, no significant association, and apparently inverse associations [13, 23, 27]. Therefore, the contribution of the present study is not the discovery of a completely new direction of association, but rather the further interpretation of this inconsistency and its potential explanations within a standardized Chinese community-based dataset. In the present study, the crude inverse association between current smoking and DED was attenuated after multivariable adjustment, whereas alcohol consumption was not independently associated with DED. Meanwhile, current smokers or drinkers showed longer tear-film stability measures but also had greater meibomian gland loss. These findings suggest that the relationship between lifestyle exposures and DED outcomes may be influenced by confounding factors, and that functional and structural ocular-surface parameters do not necessarily change in the same direction.

Accordingly, interpretation of lifestyle-related factors in DED should not rely solely on a single outcome or a single objective parameter. Apparently protective associations in observational studies may be affected by multiple factors, including population characteristics, exposure definitions, covariate structure, and discordance between ocular-surface functional and structural indicators, rather than directly reflecting true biological protection. These findings provide new community-based evidence for understanding the multifactorial nature and phenotypic heterogeneity of DED.

Public health implications

From a public health perspective, these findings caution against interpreting apparent protective associations in observational studies as evidence supporting harmful behaviors. Behavioral, environmental, and psychological factors should be considered together when translating epidemiological findings into DED prevention strategies. These findings do not support smoking or alcohol consumption as protective behaviors; smoking cessation and avoidance of excessive alcohol consumption remain appropriate public health messages.

Strengths and limitations

This study has several strengths. It was conducted within the Beijing Adult Dry Eye Cohort, a community-based cohort with standardized baseline assessment and follow-up design. Although the present analysis used only baseline cross-sectional data, the cohort provides a platform for future longitudinal evaluation of DED-related risk factors and ocular-surface changes. In addition, DED was defined using both symptoms and an objective sign of tear-film instability rather than symptom score alone. The study also incorporated multiple objective ocular-surface parameters and a broad range of demographic, environmental, lifestyle, and psychological factors.

Several limitations should also be acknowledged. The cross-sectional design precludes causal inference, and follow-up data were not included in the present analysis. Smoking and alcohol consumption were self-reported and classified only as current versus non-current exposures; therefore, information on exposure intensity, duration, cumulative dose, former use, passive smoking, alcohol type, and drinking quantity was not available. Future prospective studies with longitudinal assessment of smoking and alcohol consumption status and more detailed quantification of exposure intensity and duration are needed to better elucidate their temporal and dose-related associations with DED. Residual confounding remains possible despite adjustment for multiple covariates. In addition, the study was conducted in urban communities in Beijing and included only participants aged 45 years or older, which may limit generalizability to younger, rural, or otherwise different populations. Finally, comparisons of objective ocular-surface parameters across exposure groups were exploratory and should not be interpreted as evidence of causal effects.

Conclusions

In this community-based cohort of middle-aged and older adults, the apparent inverse association between current smoking and DED in the unadjusted analysis was no longer significant after adjustment for demographic, environmental, lifestyle, and psychological factors, and current alcohol consumption was not independently associated with DED. Current smokers and current drinkers showed longer NIBUT but also greater meibomian gland loss, suggesting that tear-film functional measures and meibomian gland structural parameters did not vary in parallel across exposure groups. These findings suggest that current smoking and current alcohol consumption should not be interpreted as beneficial for DED. Isolated favorable tear-film findings should be interpreted together with meibomian gland status and other ocular-surface features. From a public health perspective, smoking cessation and avoidance of excessive alcohol consumption remain appropriate health messages.

Abbreviations

ADEC

Beijing Adult Dry Eye Cohort

DED

Dry eye disease

OSDI

Ocular Surface Disease Index

NIBUT

Non-invasive tear break-up time

TMH

Tear meniscus height

TMA

Tear meniscus area

HAM-A

Hamilton Anxiety Scale

HAM-D

Hamilton Depression Scale

OR

Odds ratio

CI

Confidence interval

SD

Standard deviation

IQR

Interquartile range

TFOS DEWS III

Tear Film and Ocular Surface Society Dry Eye Workshop III

Authors’ contributions

YLW and KC contributed equally to this work. YLW contributed to the study design, data analysis, and manuscript drafting. KC contributed to data analysis, statistical guidance, verification of variable coding and logistic regression results, and revision of the statistical presentation in the manuscript. QZ participated in data collection and data management. JYT contributed to critical revision of the manuscript, refinement of the interpretation and discussion, language editing, and literature updating. LT and YJ contributed to study conception, supervision, and critical revision of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by Beijing Hospitals Authority’s Ascent Plan (DFL 20240202); Capital Health Research and Development of Special (Grant No. 2022-1-1081 to Dr. Jie); The Youth Beijing Scholars program (No. 022); High Level Public Health Technical Talents Construction Project from Beijing (Ying Jie); National Natural Science Foundation of China (81970764); “Yicheng Outstanding Talents” in Beijing Economic-Technological Development Area. The funders had no role in the study design, data collection, analysis, interpretation of data, or in the writing of the manuscript.

Data availability

The datasets used and/or analyzed during the current study are not publicly available due to restrictions related to participant privacy but are available from the corresponding authors on reasonable request.

Declarations

Ethics approval and consent to participate

This observational cohort study was approved by the Ethics Committee of Beijing Tongren Hospital, Capital Medical University (Ethics approval number: TREC2024-KY084), and was conducted in accordance with the principles of the Declaration of Helsinki. Signed written informed consent was obtained from each participant after explanation of the nature and possible consequences of the study. This study was registered at ClinicalTrials.gov (https://clinicaltrials.gov/). The Clinical Trial Number is NCT06621446, and the Registration Date was September 23, 2024.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.