Objective
To evaluate symptom-sign discordance in dry eye disease (DED) using 3 distinct analytical frameworks in a large prospective cohort and to identify associated systemic risk factors.
Design
Secondary analysis of a multicenter randomized clinical trial.
Participants
A total of 535 participants enrolled in the Dry Eye Assessment and Management (DREAM) study with moderate to severe DED.
Methods
Discordance was assessed at baseline using 3 approaches: (1) a rank-based comparison of symptom-sign severity; (2) a discomfort-based classification using Brief Ocular Discomfort Inventory (BODI) questionnaire; and (3) a novel temporal method evaluating symptom-sign changes between months 3 and 12. Patients were categorized as concordant, symptom-dominant, or sign-dominant discordant using each method. Multivariable regression models were used to identify predictors of discordance.
Main Outcome Measures
Prevalence and characteristics of symptom-sign discordance across methods; strength of associations between symptoms, discomfort, and clinical signs; and clinical and systemic predictors of discordance.
Results
Using the rank-based method, 77% of patients were discordant. Symptom-dominant discordance (37.8%) was associated with younger age and greater ocular discomfort, while sign-dominant discordance (39.3%) was more common in patients with Sjögren’s syndrome and was associated with greater relief from treatment. Discomfort-based analysis showed strong correlations between ocular discomfort and subjective symptoms (ρ = 0.47, p <.001), but weak associations with clinical signs. Temporal analysis revealed 46% of patients had discordant symptom-sign trajectories over 9 months, while 54% demonstrated concordance. Concordance was highest for TBUT (47.5%) and lowest for Schirmer’s test (32.7%). Antidepressant use was associated with concordance, while female sex and higher bodily pain scores were associated with symptom-dominant discordance.
Conclusions
Symptom-sign discordance in DED is common, multifactorial, and persists over time. This is the first study to apply and compare 3 discordance classification methods in a single cohort. Each offers distinct clinical utility, allowing clinicians to better stratify DED patients and enabling more targeted treatment and referral strategies, thus improving outcomes.
INTRODUCTION
D ry eye disease (DED) is a prevalent and often debilitating condition, particularly common among females. , It affects the ocular surface and is associated with symptoms such as pain, irritation, and visual disturbances, which can substantially impair quality of life. Despite its widespread occurrence, DED remains complex and multifactorial, with an incompletely understood pathophysiology. The International Dry Eye Workshop II (2017) defines DED as “a multifactorial disease of the ocular surface characterized by a loss of homeostasis of the tear film, accompanied by ocular symptoms, in which tear film instability and hyperosmolarity, ocular surface inflammation and damage, and neurosensory abnormalities play etiological roles.”
A major challenge in diagnosing and managing DED lies in the discordance between patient-reported symptoms and clinical signs. Many patients report severe discomfort despite minimal observable findings, while others exhibit significant signs with few symptoms. This mismatch complicates clinical decision-making and presents a major obstacle in research efforts aimed at understanding DED. Underlying mechanisms may include tear film instability, ocular surface inflammation, neurosensory dysfunction, and potential central sensitization. ,,,
The Dry Eye Assessment and Management (DREAM) study offers a valuable framework for examining these complexities in a structured and longitudinal manner. This large multi-center, randomized clinical trial evaluated omega-3 fatty acid supplementation for the treatment of moderate to severe DED and enrolled 535 patients across 27 clinical sites. The resulting dataset provides a robust foundation for in-depth analysis of symptom-sign discordance.
Accordingly, this study evaluated symptom-sign discordance in DED using DREAM data across 3 complementary analytical approaches, examined associated demographic and systemic factors, and explored how discordance relates to ocular discomfort and longitudinal disease course.
METHODS
Study Design and Population
The DREAM study was a multicenter, randomized clinical trial conducted across 27 clinical sites in the United States. A total of 535 participants with moderate to severe dry eye symptoms were enrolled and randomized to active or placebo supplements for 12 months. 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.
Participants completed standardized symptom questionnaires at each visit (screening visit, baseline, 3, 6, and 12 months), including the Ocular Surface Disease Index (OSDI) and Brief Ocular Discomfort Inventory (BODI), to assess symptom severity and the impact of DED on daily functioning. Comprehensive ocular surface assessments for DED signs were performed for each eye at each study visit, using 5 tests: (1) lissamine green conjunctival staining, graded on a scale of 0 to 3 based on modified National Eye Institute (NEI) guidelines, (2) corneal fluorescein staining, also graded on a 0 to 3 scale following NEI guidelines, (3) tear break-up time (TBUT), (4) Schirmer test with anesthetic, and (5) grading of Meibomian gland plugging and quality of the secretions.
This secondary analysis of DREAM study data aimed to identify symptom-sign discordance in dry eye disease and explore associated risk factors. Despite minimal clinical findings, some patients report significant dry eye symptoms, highlighting the need to better understand this subgroup. To address this, the dry eye cohort was retrospectively categorized into discordant or concordant groups using 3 different analytical methods for further evaluation ( Table 1 ).
Table 1
Definitions of Discordance Across Rank-Based, Discomfort-Based, and Temporal methods.
| Symptom-Dominant | Sign-Dominant | Concordant | |
|---|---|---|---|
| Rank-Based | Symptoms > signs (discordance score > 0) | Signs > symptoms (discordance score < 0) | Symptoms ≈ signs (discordance score ≈ 0) |
| Discomfort-Based | High discomfort level | N/A | Low discomfort level |
| Temporal | Symptoms worsen, signs unchanged or improve | Symptoms improve, signs unchanged or worsen | Both symptoms and signs changed in the same direction |
This table summarizes how discordance and concordance were defined in each analytical framework. Symptom-dominant Discordance Indicates Higher reported symptoms relative to signs; Sign-dominant discordance indicates more severe signs than symptoms.
Discordance Methods
Rank-based discordance method
A rank-based discordance score was calculated using previously described methods. , This score represented the difference between the rank-transformed OSDI symptom score and a composite severity score of signs derived from 5 objective signs: TBUT, corneal fluorescein staining, conjunctival lissamine green staining, Schirmer’s test with anesthesia, and meibomian gland assessment (based on plugging and secretion). Each test was normalized to a 0-1 scale via linear interpolation, where 0 indicated minimal severity and 1 indicated maximum severity. The average of the 5 scaled scores was used to generate an overall sign severity score for each eye, and the higher of the 2 eye scores was used to rank the patients for composite severity score of signs. The rank of OSDI and rank of composite sign severity score were each divided by the total number of patients to generate the normalized ranked severity scores (range: 0-1) of each subject. The difference between the ranking score of symptom and sign (i.e., scores of symptoms minus scores of signs) generated the discordance score ranging from − 1 to + 1.
Symptom-dominant discordance reflected symptoms more severe than signs indicating a positive score. Sign-dominant discordance reflected signs more severe than symptoms, indicating a negative score. To account for minor variations, concordance was defined as a minimal difference between symptom and sign severity, using a predefined threshold of ± 0.1. While discordance categories were defined for descriptive purposes, the discordance score was treated as a continuous variable in subsequent linear regression models to identify associated risk factors.
Ocular discomfort-discordance classification
Ocular discomfort was assessed using BODI item #3, which measures ocular discomfort severity on a scale from 0 to 100. Scores were categorized into 4 levels: none/mild (0-20), low moderate (30-40), high moderate (50-60), and severe (70-100). These strata were used to evaluate the functional burden of discomfort in dry eye disease, including its associations with SF-36 quality-of-life domains and patient-reported symptom severity.
Although discomfort level is not a formal definition of discordance, BODI #3 was selected as a surrogate marker based on prior findings from our group. In a previous study, higher average ocular pain scores over the prior week (on a 0-10 scale) were associated with increased symptom severity (higher OSDI) and milder clinical signs (lower corneal staining), consistent with discordance between symptoms and signs. These results support the use of discomfort severity as a clinically meaningful indicator of symptom-sign mismatch.
While signs were not explicitly used to define discordance categories in this method, correlations between discomfort levels and both subjective and objective measures were analyzed to evaluate the degree of alignment between reported symptoms and ocular surface findings.
Temporal discordance analysis
Discordance was defined as a mismatch in the direction of change between signs and symptoms over a 9-month period. Symptom and sign data were obtained from month 3 (M3) and month 12 (M12), with M3 selected as the baseline to mitigate early placebo effects observed in the DREAM study. Specifically, symptom-dominant discordance occurred when symptoms worsened while signs improved or remained stable. Sign-dominant discordance occurred when symptoms improved while signs worsened or remained stable (i.e., change in opposite directions, resulting in a negative direction). Concordance was defined as symptoms and signs changing in the same direction (i.e., positive direction), whether improving, worsening, or remaining stable ( Table 1 ). This approach was designed to capture concordance or discordance in the direction of change rather than the magnitude of response.
The direction-based temporal discordance was determined by comparing the direction of change in OSDI with the direction of change in the composite severity score of signs, derived from a combination of all 5 signs, as well as with each individual sign independently. Associations between discordance categories and various risk factors were determined to better understand potential contributors to symptom-sign mismatch in DED.
Statistical Analysis
All analyses were conducted using SAS v9.4 (SAS Institute Inc., Cary, NC). Linear regression models were used in the rank-based method to examine associations between the continuous discordance score and baseline demographic, ocular, and systemic variables. Sex was defined as a biological variable (female or male) as documented in the medical records. Spearman correlation coefficients were calculated to evaluate the relationship between ocular discomfort and both DED symptom and sign measures. Ordinal logistic regression was performed to assess associations between pain severity categories and quality-of-life outcomes. For the temporal analysis, logistic regression models were used to identify predictors of concordance (yes/no) and negative discordance (yes/no) classifications, based on both composite and individual signs. A p -value <.05 was considered statistically significant for all analyses.
RESULTS
Baseline Characteristics
A total of 535 patients were included in the study. The mean age was 58.0 ± 13.2 years, and the majority were female (81%). Most participants identified as White (74%), followed by Black (12%) and other racial groups (14%). Thirteen percent of the cohort identified as Hispanic or Latino.
At baseline, the mean (SD) total OSDI score was 42.1 ± 15.5. Subscale scores included vision-related function (35.2 ± 19.2), ocular symptoms (45.1 ± 19.1), and environmental triggers (52.6 ± 26.8). Mean (SD) for clinical signs included conjunctival staining score (2.95 ± 1.41), corneal fluorescein staining score (3.80 ± 2.81), TBUT (3.14 ± 1.58 seconds), Schirmer’s test results (9.57 ± 6.71 mm/5 min), and meibomian gland abnormality (3.09 ± 1.70).
Rank-Based Discordance
The distribution of discordance scores at baseline is shown in Figure 1 . Scores ranged from − 1 to + 1, with a near-normal distribution centered around zero, indicating that most participants exhibited relatively balanced symptom-sign relationships. Using a predefined threshold of ± 0.1 to define concordance, 202 patients (37.8%) were classified as having symptom-dominant discordance, 210 patients (39.3%) as sign-dominant discordance, and 123 patients (23.0%) as concordant.
Baseline distribution of symptom-sign discordance scores using the rank-based method. The histogram shows the distribution of discordance scores in 535 DREAM study participants, calculated by subtracting the composite rank of clinical signs from the OSDI symptom score. Values near zero indicate concordance, while positive and negative values reflect symptom- and sign-dominant discordance, respectively. A threshold of ± 0.1 was used to classify concordant cases. Most participants demonstrated some degree of discordance, highlighting the limited alignment between patient-reported symptoms and objective findings in dry eye disease.
Univariate linear regression was performed to explore associations between baseline characteristics and discordance score ( Table 2 ). Increasing age was significantly associated with sign-dominant discordance (β = −0.004, p =.003), suggesting that older patients tended to exhibit more severe signs relative to symptoms. Higher self-reported ocular discomfort, as measured by BODI discomfort subscale (β = 0.006, P <.001) and BODI item #3 (β = 0.005, P <.001), was significantly associated with symptom-dominant discordance. Greater relief from treatment (BODI #12) was associated with sign-dominant discordance (β = −0.002, P =.007). Among systemic comorbidities, patients with Sjogren’s syndrome were more likely to demonstrate sign-dominant discordance (β = −0.175, P =.002). Other comorbidities such as rheumatoid arthritis and osteoarthritis showed similar trends but did not reach statistical significance. No meaningful associations were observed with sex, race, smoking status, depression, or antidepressant use.
TABLE 2
Univariate Analysis of Baseline Factors with Symptom-Sign Discordance Using Rank-Based Method.
| Variable | Overall ( N = 535) | Mean Symptom Rank | Mean Sign Rank | Beta | P -value |
|---|---|---|---|---|---|
| Age (years) | 58.0 (13.2) | n/a | n/a | −0.004 | 0.003 |
| General health perception | 67.9 (24.9) | n/a | n/a | REF | 0.63 |
| Bodily Pain | 47.3 (9.7) | n/a | n/a | -REF | 0.81 |
| Social Functioning | 49.6 (9.6) | n/a | n/a | −0.001 | 0.46 |
| BODI discomfort subscale | 41.7 (18.6) | n/a | n/a | 0.006 | <0.001 |
| BODI #3 average discomfort | 42.8 (19.1) | n/a | n/a | 0.005 | <0.001 |
| BODI #12 relief from treatment | 38.5 (27.5) | n/a | n/a | −0.002 | 0.007 |
| Sex | |||||
| Male | 101 (18.9%) | 0.45 | 0.41 | REF | |
| Female | 434 (81.1%) | 0.51 | 0.52 | −0.055 | 0.20 |
| Ethnicity | |||||
| Hispanic or Latino | 68 (12.7%) | 0.50 | 0.47 | REF | |
| Other | 467 (87.3%) | 0.50 | 0.50 | −0.032 | 0.53 |
| Race | |||||
| White | 398 (74.4%) | 0.51 | 0.51 | REF | |
| Black | 64 (12.0%) | 0.42 | 0.42 | 0.002 | 0.97 |
| Other | 73 (13.6%) | 0.52 | 0.50 | 0.028 | 0.58 |
| Cigarette smoking | |||||
| Never | 367 (68.6%) | 0.50 | 0.49 | REF | |
| Former | 142 (26.5%) | 0.51 | 0.55 | −0.049 | 0.20 |
| Current | 26 (4.9%) | 0.54 | 0.47 | 0.055 | 0.49 |
| Rheumatoid arthritis | |||||
| No | 486 (90.8%) | 0.50 | 0.49 | REF | |
| Yes | 49 (9.2%) | 0.50 | 0.59 | −0.105 | 0.07 |
| Osteoarthritis | |||||
| No | 394 (73.6%) | 0.51 | 0.49 | REF | |
| Yes | 141 (26.4%) | 0.48 | 0.53 | −0.071 | 0.07 |
| Diabetes | |||||
| No | 473 (88.4%) | 0.50 | 0.50 | REF | |
| Yes | 62 (11.6%) | 0.50 | 0.47 | 0.035 | 0.51 |
| Depression | |||||
| No | 413 (77.2%) | 0.49 | 0.49 | REF | |
| Yes | 122 (22.8%) | 0.53 | 0.55 | −0.025 | 0.53 |
| Fibromyalgia | |||||
| No | 523 (97.8%) | 0.50 | 0.50 | REF | |
| Yes | 12 (2.2%) | 0.55 | 0.54 | 0.011 | 0.92 |
| Anti-depressants | |||||
| No | 370 (69.2%) | 0.49 | 0.48 | REF | |
| Yes | 165 (30.8%) | 0.53 | 0.55 | −0.027 | 0.46 |
| Taking anti-depressants | |||||
| No | 406 (75.9%) | 0.49 | 0.49 | REF | |
| Yes | 129 (24.1%) | 0.52 | 0.52 | −0.002 | 0.95 |
| Sjögren’s syndrome | |||||
| No | 456 (89.8%) | 0.49 | 0.48 | REF | |
| Yes | 52 (10.2%) | 0.51 | 0.68 | −0.175 | 0.002 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
