Longitudinal Examination of Frequency of and Risk Factors for Severe Dry Eye Symptoms in US Veterans

Key Points

Questions  What are the nature of symptom progression in dry eye and the risk factors associated with severe symptoms at 1 year?

Findings  In this longitudinal study of US veterans, most patients with severe baseline dry eye symptoms reported persistent severe symptoms at 1 year. Baseline nociceptive and neuropathic ocular pain, in addition to sleep disturbances, mental health status, nonocular pain, and medications were also risk factors.

Meaning  These findings suggest that severe dry eye symptoms, baseline ocular pain, and nonocular risk factors that affect pain perception and severity are associated with severe dry eye symptoms 1 year after initial evaluation.

Importance  Dry eye syndrome is a common condition that affects millions of individuals. Many cross-sectional studies have evaluated risk factors for dry eye severity, but few have assessed risk factors or symptom progression over time.

Objectives  To assess symptom progression in dry eye syndrome and determine risk factors associated with severe symptoms at 1 year.

Design, Setting, and Participants  A longitudinal study was conducted from October 1, 2013, to April 30, 2015, among patients at the Miami Veterans Affairs Hospital with a wide variety of dry eye symptoms and signs (ranging from none to severe).

Main Outcomes and Measures  Change in dry eye symptom severity during 1 year, as assessed by responses to dry eye symptom questionnaires administered at the initial visit and 1 year later, as well as baseline risk factor analysis for severe dry eye symptoms at 1 year, defined as a Dry Eye Questionnaire 5 score of 12 or more.

Results  Of the 120 patients (mean [SD] age, 64 [11] years; 109 male and 11 female), 26 of 58 (44.8%) with either no symptoms or mild or moderate symptoms at baseline progressed to more severe symptoms at 1 year, while 46 of 62 patients (74.2%) with severe symptoms at baseline reported that severe symptoms persisted at 1 year. Baseline ocular risk factors for severe dry eye symptoms at 1 year included more severe dry eye symptoms, ocular pain, and neuropathic pain–like ocular symptoms. Nonocular risk factors included sleep disturbances (eg, sleep apnea and insomnia), mental health status (eg, posttraumatic stress disorder and depression), nonocular pain, and medications (eg, anxiolytics and analgesics). In a multivariable analysis, the most significant risk factors were sleep apnea (odds ratio [OR], 3.80; 95% CI, 1.00-14.49; P = .05), Dry Eye Questionnaire 5 score (OR, 1.15; 95% CI, 1.02-1.30; P = .02), and posttraumatic stress disorder score (OR, 1.04; 95% CI, 1.01-1.08; P = .02).

Conclusions and Relevance  Patients with severe dry eye symptoms and ocular pain at baseline were more likely to have persistent severe dry eye symptoms on 1-year follow-up. Furthermore, nonocular risk factors that have been associated with dry eye cross-sectionally, such as psychiatric comorbidities and nonocular pain, were also associated with severe dry eye symptoms at 1 year. Although this cohort was limited to US veterans, which may not be generalizable to other populations, our results suggest that pain perception and severity are important when evaluating and managing dry eye.

Quiz Ref IDDry eye is a common condition that affects the quality of life of millions of people worldwide.¹ The International Dry Eye Workshop report estimates the prevalence of dry eye to range from 5% to 30% of persons 50 years or older, and this number will likely increase as our population ages.² Symptoms of dry eye can prompt frequent physician visits and negatively affect quality of life.^3-8 Common symptoms include pain, often characterized by patients as dryness or burning, and visual changes, such as blurry or fluctuating vision.^3,4 There are many signs of dry eye, which can include increased osmolarity,⁹ decreased tear production,¹⁰ increased tear evaporation,¹¹ and increased ocular staining.¹²

Many cross-sectional studies have demonstrated various risk factors associated with dry eye, which range from demographics (eg, increasing age^1,13-17 and female sex^13,18-20) to certain medical conditions (eg, posttraumatic stress disorder [PTSD], depression, and thyroid disease)^21-23 and the use of certain medications (eg, antidepressants²⁴ and anxiolytics^21,23). Fewer epidemiologic studies have evaluated dry eye in a longitudinal fashion. There are few data on the frequency of and risk factors for persistent dry eye symptoms 1 or more years after diagnosis. One study by Lienert et al²⁵ described the longitudinal course of dry eye assessed via a dry eye–change questionnaire during a 1-year period. The study demonstrated that ocular surface symptoms, vision-associated symptoms, and worsening of social function (eg, work satisfaction, ability to socialize, satisfaction with socializing, overall mood, irritableness, marriage quality, friendship quality, and overall health) was reported in 24%, 29%, and 10% of patients, respectively. Patients with severe dry eye symptoms at baseline had a 2-fold increased risk of worsening symptoms at 1 year. The frequency of worsening symptoms was not different by type of dry eye treatment. This finding is in contrast to a randomized study of artificial tears vs cyclosporine, 0.05%, where 32% of 22 patients treated with artificial tears had disease progression (composite of symptoms and signs) at 1 year vs 6% of 36 patients treated with cyclosporine, 0.05%.²⁶

A previous study demonstrated that patients with neuropathic pain–like ocular symptoms (hot-burning ocular pain and sensitivity to light and wind) had a more persistent and severe course of dry eye symptoms.²⁷ A limitation of that study, however, was that these pain-specific descriptors were obtained at follow-up and not at the baseline examination. As such, we could not evaluate whether these factors imparted prognostic information on the disease course. This topic is important to clinicians, as identifying patients with a more chronic disease course can alter treatment algorithms, with more aggressive intervention considered in those more likely to have a chronic course. To fill these gaps and better understand the longitudinal nature of dry eye progression, in this study, we assessed symptom severity at baseline and 1-year follow-up, and evaluated which baseline risk factors, including demographics, comorbidities, and specific ocular pain symptoms, were associated with severe dry eye symptoms at 1 year (Dry Eye Questionnaire 5 [DEQ5] score ≥12).²⁷ We hypothesized that patients with neuropathic-like ocular pain symptoms would be more likely to have severe dry eye symptoms 1 year after diagnosis.

Quiz Ref IDPatients with no overt eyelid or corneal abnormalities were prospectively recruited from the Miami Veterans Affairs Healthcare System eye clinic between October 1, 2013, and April 30, 2015, and underwent a complete ocular surface examination. As our goal was to include patients with idiopathic dry eye (ie, patients whose dry eye symptoms were without an obvious etiologic cause), we excluded patients who wore contact lenses, underwent refractive surgery, used ocular medications with the exception of artificial tears (eg, glaucoma medications), had an active external ocular process, had cataract surgery within the last 6 months, or underwent any glaucoma or retinal surgery in the past. In addition, we excluded patients with human immunodeficiency virus, sarcoidosis, graft-vs-host disease, or a collagen vascular disease. Given our inclusion and exclusion criteria, our study population included patients with and without dry eye. The Miami Veterans Administrations Medical Center Institutional Review Board (Miami, Florida) approved this study, all participants completed written informed consent, all work was compliant with the Health Insurance Portability and Accountability Act of 1996, and research adhered to the Declaration of Helsinki.

For each participant, demographic information, ocular and medical history, and medication information were collected by self-report and verified by electronic medical records. Medications were classified by categories (eg, antidepressants and anxiolytics) based on their marketed indication. Patients completed standardized questionnaires regarding dry eye symptoms, including the DEQ5 (total score, 0-22)²⁸ and the Ocular Surface Disease Index (total score, 0-100).²⁹ In addition, the participants were asked to rate the intensity of their average eye pain during a 1-week recall period using a numerical rating scale, where 0 indicated no pain sensation and 10 indicated the most intense eye pain imaginable. The Neuropathic Pain Symptom Inventory,³⁰ modified for the eye (NPSI-Eye), was administered to evaluate the severity of ocular neuropathic pain–like symptoms (total score, 0-100). In our modified version, we replaced the original items regarding allodynia and hyperalgesia (pain caused by brushing, pressure, or cold on the skin; questions 8, 9, and 10, respectively), with items specific to ocular hyperalgesia and allodynia (eye pain evoked or worsened by wind, light, and heat or cold) (NPSI-Eye). Psychological status was assessed using the 9-item version of the Patient Health Questionnaire for depression (total score, 0-27),³¹ the PTSD Checklist–Military Version (total score, 17-85),³² and the Symptom Checklist-90 for anxiety (total score, 0-4).³³ Insomnia was assessed with the Insomnia Severity Index, a brief, 7-item instrument measuring the patient’s perception of his or her insomnia (total score, 0-28).³⁴

All patients underwent tear film assessment, including, in the order performed, measurement of (1) tear osmolarity once in each eye (TearLAB), (2) InflammaDry testing (Rapid Pathogen Screening Inc) once in each eye, and (3) mechanical detection of pain thresholds of the right central cornea via modified Belmonte noncontact aesthesiometer.³⁵ After a 2-hour break, the examination continued with tear breakup time (5 µL of fluorescein placed, 3 measurements taken in each eye and averaged) and measurement of the presence and location of conjunctivochalasis, defined as an absent tear-lake with the replacement of the cul-du-sac with conjunctival tissue. The locations of these conjunctival folds (nasal, middle, or temporal) were recorded. The examination continued with corneal epithelial cell disruption measured via corneal staining (National Eye Institute scale,³⁶ 5 areas of cornea assessed; score 0-3 in each area, total 15), testing of tear production via Schirmer strips with anesthesia, measurement of eyelid vascularity (0, none; 1, mild; 2, moderate; 3, severe engorgement), assessment of meibomian gland dropout via meibography and graded to the Meiboscale (total score, 0-4),³⁷ and determination of meibum quality (0, clear; 1, cloudy; 2, granular; 3, toothpaste; 4, no meibum extracted). All ocular surface evaluations were performed for both eyes in each patient, with only the poorest of the 2 scores included for statistical analysis.

One year after study enrollment, patients were contacted by telephone and asked about use of artificial tears and other medication, and again administered the DEQ5 and NPSI-Eye. Of 296 individuals, 120 completed the 1-year follow-up questions; 9 did not wish to complete the survey, 166 could not be reached by telephone, and 1 patient was deceased. Demographics, comorbidities, and baseline dry eye symptoms and signs were similar between participants and nonparticipants. The 2 outcome measures were change in severity of dry eye symptoms during a 1-year period and baseline risk factor analysis for severe dry eye symptoms at 1 year (DEQ5 score of ≥12).

All statistical analyses were performed using SPSS, version 22.0 (SPSS Inc). Descriptive and frequency statistics were used to summarize patient baseline demographics and clinical information. Univariate logistic regression analysis was used to assess correlations between baseline factors and DEQ5 scores of 12 or more at 1-year follow-up. Using multivariable logistic analysis via forward stepwise logistic regression, we then evaluated which factors were most robustly associated with dry eye symptoms at 1 year. We obtained odds ratios (ORs) and corresponding 95% CIs for each regression. The sample size was deemed appropriate to detect factors with moderate effect size. For symptom persistence, for example, with a sample size of 120 patients, response probability of 15%, and significance set at P = .05, the power was greater than 90% to demonstrate effects of factors with ORs of 2.0 or more in a multiple logistic regression model for many types of variables (ie, normally distributed continuous, dichotomous, ordinal).

A total of 120 patients participated in the study (mean [SD] age, 64 [11] years; 109 men and 11 women). Demographics, comorbidities, and medication use by the study population are presented in Table 1.

Of 120 patients, 14 (11.7%) presented with no dry eye symptoms (DEQ5 score ≤6) at baseline. Of those, 7 patients (50.0%) remained symptom free, while 3 (21.4%) progressed to mild or moderate symptoms (DEQ5 score 6-11) and 4 (28.6%) progressed to severe symptoms (DEQ5 score ≥12). Another 44 patients (36.7%) had mild or moderate symptoms at baseline; in a similar manner, 22 of these patients (50.0%) progressed to severe symptoms at 1 year, while 11 (25.0%) had stable symptoms and 11 (25.0%) had no symptoms. Of the 62 patients (51.7%) with severe symptoms at baseline, most (46 [74.2%]) had persistent severe symptoms, while 11 (17.7%) regressed to mild or moderate symptoms and 5 (8.1%) to no symptoms.

Using univariable logistic regression analysis, we examined which demographics, comorbidities, and medications were risk factors for severe dry eye symptoms (DEQ5 score, ≥12) at 1 year. Having a diagnosis of sleep apnea increased the risk approximately 2.5-fold (OR, 2.63; 95% CI, 1.02-6.75; P = .045). More severe PTSD (assessed via PTSD questionnaire) (OR, 1.03; 95% CI, 1.01-1.06; P = .005), depression (assessed via the 9-item Patient Health Questionnaire) (OR, 1.09; 95% CI, 1.02-1.16; P = .006), insomnia (assessed via the Insomnia Severity Index) (OR, 1.08; 95% CI, 1.01-1.15; P = .03), and nonocular pain (mean pain assessed via 1-week recall) (OR, 1.23; 95% CI, 1.06-1.41; P = .005) also increased the risk of severe symptoms at 1 year (Table 1). Antianxiety medications (OR, 3.23; 95% CI, 1.40-7.44; P = .006) and analgesics (OR, 2.53; 95% CI, 1.19-5.38; P = .02) were also risk factors for more severe symptoms at 1 year.

Quiz Ref IDWe next examined baseline dry eye metrics and evaluated which metrics were associated with severe dry eye symptoms at 1 year. Individuals with more severe dry eye symptoms (assessed via the DEQ5 [OR, 1.18; 95% CI, 1.02-1.28; P < .001] and Ocular Surface Disease Index [OR, 1.02; 95% CI, 1.01-1.04; P = .008]) and ocular pain (assessed via the numerical rating scale [OR, 1.22; 95% CI, 1.05-1.42; P = .01]) at baseline were more likely to have severe dry eye symptoms at 1 year (Table 2). Similarly, individuals with neuropathic pain–like ocular symptoms, as assessed by the NPSI-Eye at initial visit, were also at risk of severe symptoms at 1 year (total score: OR, 1.04; 95% CI, 1.01-1.06; P = .002). On the other hand, none of the dry eye signs examined correlated with symptom severity at 1 year.

To test the robustness of the association between baseline metrics and severe dry eye symptoms at 1 year, we performed a multivariable analysis of demographics (age, sex, and race or ethnicity), medication use (antidepressants, analgesics, gabapentin), mental health (depression, PTSD), medical comorbidities (sleep apnea, insomnia), and dry eye symptoms at baseline (assessed via the DEQ5, NPSI-Eye, and Ocular Surface Disease Index). In this model, sleep apnea was the most important risk factor for severe dry eye symptoms at 1 year (OR, 3.80; 95% CI, 1.00-14.49; P = .05), followed by the DEQ5 score (OR, 1.15; 95% CI, 1.02-1.30; P = .02) and PTSD score (OR, 1.04; 95% CI, 1.01-1.08; P = .02).

Quiz Ref IDDespite the major prevalence of the disease, this study is one of the few longitudinal reports on dry eye, to our knowledge. Results demonstrated that dry eye symptoms are not static and that approximately 50% of patients with no, mild, or moderate symptoms progressed to severe symptoms at 1 year. On the other hand, most patients with severe symptoms at baseline had persistent, severe symptoms at 1 year. This finding is in line with the finding in the study by Lienert et al²⁵ that evaluated self-reported change in dry eye 1 year after initial evaluation. In their population of patients with dry eye, with a mean disease duration of 10.5 years in men and 14.5 years in women, after 1 year, worsening ocular surface symptoms were noted in 24% of patients, vision-associated symptoms in 29%, and worsening of social function (eg, work satisfaction, ability to socialize, satisfaction with socializing, overall mood, irritableness, marriage quality, friendship quality, and overall health) in 10%. Consistent with our study, the most reliable metric of disease worsening was a history of severe symptoms, reinforcing the tendency of baseline symptom severity to indicate disease chronicity. However, the overall median change in disease score for the Lienert et al²⁵ study was 0, suggesting that symptom progression is not uniform.

Another objective of our study was to define risk factors for severity of dry eye 1 year after initial evaluation, hypothesizing that neuropathic-like ocular pain symptoms would be a risk factor, as neuropathic pain elsewhere tends to be chronic.^38,39 Overall, we found that some risk factors associated with severity of dry eye at 1 year were similar to those identified in previous cross-sectional studies, such as sleep apnea,^21,40 depression,^21,41-43 PTSD,^21,43,44 and use of antianxiety medication.^21,23 Unlike other studies,^23,24 however, we did not find an association with antidepressants. Specific to our study, we found that neuropathic-like ocular pain symptoms (as assessed by the NPSI-Eye) were risk factors for severe dry eye symptoms at 1 year. This finding reinforces those of a previous study where the same neuropathic-like ocular pain symptoms measured 2 to 3 years after initial evaluation correlated with severity and persistence of dry eye symptoms.²⁷ These studies suggest that the NPSI-Eye may aid in determining chronicity and severity of dry eye symptoms.

Given the lack of a criterion standard diagnosis for neuropathic ocular pain, our validation strategy for the NPSI-Eye has been in studying criterion validity; that is, asking whether the test models what is known about neuropathic pain.⁴⁵ With this framework in mind, we found that total and subscale scores of the NPSI-Eye in our population were similar to those found in other cohorts with neuropathic pain,⁴⁶ and specific descriptors within the NPSI-Eye questionnaire (spontaneous burning pain, sensitivity to light, and sensitivity to wind) were associated with a chronic disease course and a decreased response to local therapy.^27,47 These findings lend support to the questionnaire’s usefulness in identifying a more severe subtype of dry eye that may be driven by underlying somatosensory dysfunction. Questions regarding ocular burning and sensitivity to wind and light also have been important in assessment of ocular pain over time using the Ocular Pain Assessment Survey.⁴⁸

There is biological plausibility to explain the development of neuropathic pain within the eye. The corneal nociceptors are in close proximity to the ocular surface and can be damaged by trauma (eg, surgery), environmental stress (eg, air pollution), and tear film abnormalities (eg, hyperosmolarity). Repeated stress can lead to peripheral sensitization, which may subsequently lead to changes in the central nervous system.⁴⁹ Cross-sectional studies have found increased pain sensitivity and enhanced temporal summation at a site remote from the eye (forearm) in patients with dry eye symptoms, supporting pain sensitization as a possible explanation for more severe symptoms of dry eye.^50,51 This association is further reinforced by cross-sectional studies reporting increased severity of dry eye symptoms in those with self-reported chronic pain syndromes, even in the absence of ocular surface signs.^52,53

There is also biological plausibility to support the noted association between nonocular comorbidities (sleep apnea, mental health, and pain intensity) and dry eye severity at 1 year. Sleep apnea is associated with eyelid laxity,^54,55 which has been found to be a risk factor for severe ocular surface symptoms,⁵⁶ perhaps owing to disruption of the healthy tear film dynamic.⁵⁷ Mental health conditions may either predispose individuals to pain or be a result of pain.^58,59 For example, depression was repeatedly identified as a risk factor for a variety of pain disorders in many studies.⁵⁸ Likewise, a multicenter study involving patients with at least a 3-month history of peripheral neuropathy found that anxiety and mood disorders were extremely prevalent,⁶⁰ perhaps owing to an inability to cognitively cope with pain.⁶¹ Consistent with other studies,^62-64 they also found a strong connection between insomnia and pain intensity, with two-thirds of patients reporting sleep interference.⁶⁰ Similar to dry eye, severity of nonocular pain has been found to be associated with pain persistence in surgical models.^65,66 A prospective study in the Netherlands found that a high acute postoperative pain intensity score was an important metric of chronic pain 12 months after hysterectomy.⁶⁷ Neuropathic pain was also found to be a risk factor for pain chronicity, with 53% of individuals with chronic pain after surgery reporting neuropathic pain symptoms 4 days after surgery vs 19% in a group without chronic pain.

This study has limitations that must be considered when interpreting its findings. First, not all patients completed the 1-year evaluation. For most patients who did not complete the evaluation, they were unable to be reached by telephone. A few patients refused to complete the questionnaire owing to time constraints. However, demographic characteristics were similar between responders and nonresponders. Second, a repeated ocular surface examination was not performed and, as such, we cannot comment on the stability of ocular surface signs at 1 year. We chose to focus on dry eye symptoms, as these are the most bothersome component of disease and previous studies have demonstrated a disconnect between symptoms and peripheral signs of disease.^68,69 In addition, only 1 follow-up time point was assessed; thus, fluctuation of symptoms over time cannot be addressed. Following up at 6 months would have provided additional data points as well as more information on the value of reviewing patients’ dry eye symptoms more frequently than at 1 year. Finally, the cohort was limited to mostly male US veterans in Miami, Florida, which may not be generalizable to other populations with dry eye, although it is encouraging that our population had similar risk factors for dry eye as did other diverse cross-sectional populations.^23,24

This study highlights the important role of pain severity and perception on severity and persistence of dry eye. Our data suggest that dry eye questionnaires should be expanded to include questions about the time course of dry eye symptoms, pain severity, and neuropathic qualities. These questionnaires can be easily and inexpensively incorporated into a busy clinical practice (ie, provided by front desk staff or technicians) and made available for review by the time the patient is ready to be seen. In addition, short and standardized questionnaires for depression and anxiety can help identify patients who may benefit from a referral to a mental health care professional. Additional longitudinal studies are needed to characterize dry eye findings beyond 1 year and evaluate which pain management strategies, whether systemic or local, will be most beneficial in patients with dry eye.

Accepted for Publication: October 25, 2016.

Corresponding Author: Anat Galor, MD, MSPH, Bascom Palmer Eye Institute, University of Miami, 900 NW 17th St, Miami, FL 33136 (agalor@med.miami.edu).

Published Online: December 22, 2016. doi:10.1001/jamaophthalmol.2016.4925

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

Study concept and design: Alghamdi, Levitt, Sarantopoulos, Felix, Galor.

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

Drafting of the manuscript: Ong, Alghamdi, Levitt, Sarantopoulos, Galor.

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

Statistical analysis: Ong, Alghamdi, Galor.

Obtained funding: Levitt.

Administrative, technical, or material support: Alghamdi, McClellan, Felix.

Study supervision: Alghamdi, Sarantopoulos, Felix.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This study was supported by grant EPID-006-15S from the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Clinical Sciences Research (Dr Galor), Center Core Grant P30EY014801 from the National Institutes of Health, and an unrestricted grant from Research to Prevent Blindness.

Role of Funder/Sponsor: The funding sources 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.

Additional Information: The Belmonte noncontact aesthesiometer was built by the Brien Holden Vision Institute in Australia and modified by Jean Marie Parel at Bascom Palmar. It is not a commercially available device.