Article Text

Prevalence and risk factors of dry eye disease in a British female cohort
  1. Jelle Vehof1,2,
  2. Diana Kozareva1,
  3. Pirro G Hysi1,
  4. Christopher J Hammond1,3
  1. 1Department of Twin Research & Genetic Epidemiology, King's College London, St Thomas’ Hospital, London, UK
  2. 2Department of Ophthalmology & Epidemiology, University Medical Center Groningen, Groningen, The Netherlands
  3. 3Department of Ophthalmology, King's College London, St Thomas’ Hospital, London, UK
  1. Correspondence to Dr Jelle Vehof, Department of Twin Research & Genetic Epidemiology, King's College London, St Thomas’ Hospital, Westminster Bridge Road, London SE1 7EH, UK; j.vehof{at}


Background/aims To estimate the prevalence and risk factors of dry eye disease (DED) in a female cohort in the UK.

Methods Population-based cross-sectional association study of 3824 women from the TwinsUK cohort aged 20–87 years. A questionnaire was used to evaluate DED and several risk factors. Binary logistic regression, corrected for age, was used to examine the association between DED and risk factors.

Results 9.6% of women had a DED diagnosis and concomitant use of artificial tears, and 20.8% experienced DED symptoms in the past 3 months. Risk factors that were significantly associated with DED were age, asthma, eczema, the presence of any allergy, cataract surgery, rheumatoid arthritis, osteoarthritis, migraine and stroke. The highest effect sizes were found with depression, pelvic pain, irritable bowel syndrome and chronic widespread pain syndrome (all p<0.0005). Subjects with DED symptoms scored significantly lower on self-perceived health, compared with controls (p=0.001).

Conclusions DED is common and increases with age within this cohort of female twins. We confirmed established risk factors for the first time in a British population, and found important risk factors that might relate to an underlying aetiology involving chronic pain predisposition or somatisation.

  • Epidemiology
  • Cornea
  • Conjunctiva
  • Ocular surface

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Dry eye disease (DED) is a multifactorial disease of the tears and the ocular surface, which results in symptoms of discomfort, visual disturbance, and tear film instability with potential damage to the ocular surface.1 It is a serious health problem, with a prevalence of 7–34% in middle-aged and elderly subjects, depending on the country, population and definition used.2 The pathophysiology and aetiology of DED are poorly understood and most likely multiple mechanisms are involved.3 No generally agreed gold standard for the diagnosis of DED exists and treatment options are limited.

Studies searching for risk factors of DED are important, because they provide clues for pathophysiological mechanisms and give potential treatment options by excluding the risk factor where possible. Several risk factors have repeatedly been associated with DED in clinic-based case-control and population-based epidemiology cohorts, for example, increasing age, contact lens wear and certain medications, such as antihistamines and antidepressants.2 ,4 Women have most repeatedly been found to be at increased risk compared with men, particularly during or after the menopause.2

To date, there have been few population-based epidemiological studies on DED and none in the UK. The aim of this study is to estimate the prevalence and risk factors of DED in a female population-representative cohort in the UK. As well as attempting to replicate findings from other studies (to establish a consensus on known risk factors, given the paucity of studies), several new hypotheses for DED were tested, such as the association with other chronic pain disorders.

Material and methods

Study population

The local Research Ethics Committee approved all procedures, and all subjects provided informed consent. The study was conducted on 3824 female twin volunteers from 2410 families (1414 complete twin pairs and 996 unrelated twin subjects), aged 20–87 years, from the TwinsUK adult registry held at St Thomas’ Hospital, London in 2012. This is a cohort of largely female twins that has been recruited from the general population via successive local and national media campaigns.5 The TwinsUK cohort is representative of the general population and has been found to be comparable to a UK singleton female cohort for a wide range of diseases and traits.6 It has been used in many epidemiological studies of healthy ageing.7 Although subjects were part of a twin pair, we did not make use of the twin relationship in this study.


DED cases were defined as having a diagnosis of DED made by a clinician and concomitant treatment with artificial tears, as used in other population-based studies.8 This was assessed by the following two questions: (1) “Have you ever been diagnosed (by a clinician) as having dry eye syndrome?”, (2) “Do you currently use artificial tear eye drops or gel?”, If a participant answered ‘Yes’ to both questions, she was classified as a case of DED (further also regarded as ‘having a DED diagnosis’). If a participant answered ‘No’ to both, she was classified as a control. When only one of the two questions was positive, subjects were excluded from analyses. A third question was used as a proxy for DED symptoms,9 independent from a diagnosis or therapy: (3) “For the past 3 months or longer, have you had dry eyes? (This is described as a foreign body sensation with itching and burning, sandy feeling, not related to allergy)”. Participants who answered ‘Do not know’ were excluded from analyses for that question.

Potential risk factors

Structured postal questionnaires were administered to all participants. They included basic demographic data, and a past history of systemic and ophthalmological diseases by using the question “Has a doctor ever diagnosed or treated you for any of the following conditions?”.

Self-perceived health and OSDI

A random subset of 681 persons who participated in a substudy on pain (see Ref. 10) completed additional questionnaires on dry eye symptoms and self-perceived health. Self-perceived health was rated on a 5-point ordinal scale from poor (1) to excellent (5). This same subset completed the Ocular Surface Disease Index (OSDI). The OSDI, developed by the Outcomes Research Group at Allergan (Irvine, California), is a 12-item questionnaire designed to provide a rapid assessment of the symptoms of ocular irritation consistent with DED and their impact on vision-related functioning. Each answer is scored on a 5-point scale, leading to a total OSDI score ranging from 0 (no symptoms) to 100 (maximum symptoms).11

Statistical analysis

All statistical analyses were performed using SPSS V.20.0 (SPSS, Chicago, Illinois, USA) statistical package. Analyses of DED were stratified in (1) DED cases versus controls and (2) subjects with dry eye symptoms versus subjects without dry eye symptoms. Prevalence was calculated within each decade of age and binary logistic regression was used to assess the relationship between DED and age. Binary logistic regression corrected for age was used to examine associations between DED and other potential risk factors. Linear regression corrected for age was used to examine associations between self-perceived health and DED symptoms. OR and 95% CIs were calculated. A two-sided p value of 0.05 or lower was considered statistically significant for all analyses.


Characteristics of the study population are shown in table 1. Mean age of the complete study population was 57.1 (SD 13.1) years. The overall prevalence of DED (diagnosed by a clinician and using artificial tear eye drops or gel) was 9.6% (95% CI 8.7% to 10.6%) and the prevalence of DED symptoms in the past 3 months was 20.8% (95% CI 19.5% to 22.1%). The prevalence of DED diagnosis and dry eye symptoms by age group are shown in figure 1. DED diagnosis prevalence rose with age from 2.7% in the third decade to 20.0% in the ninth decade (OR of 1.045 (95% CI 1.035 to 1.055) per year, p<0.0005). Also, dry eye symptoms were significantly more prevalent with increasing age (OR of 1.014 (95% CI 1.008 to 1.021) per year, p<0.0005), but symptoms were already highly present in persons in the third decade (20.3%). No differences between monozygotic and dizygotic twins in the prevalence of DED were found (Fisher's exact test, p values of 0.91 and 0.90 for DED diagnosis and dry eye symptoms, respectively).

Table 1

Characteristics of the study population

Figure 1

Prevalence (and 95% CI) among age decades of (dashed line) a diagnosis of dry eye disease with current use of artificial tears and (full line) dry eye symptoms for the past 3 months or longer.

Table 2 shows the binary logistic regression analysis (corrected for age) of the association between DED diagnosis and dry eye symptoms and potential risk factors. The presence of cataract surgery, asthma, eczema, allergies, thyroid problems, osteoarthritis, rheumatoid arthritis, fertility problems, migraine, depression, irritable bowel syndrome (IBS), pelvic pain and chronic widespread pain syndrome (CWPS) all showed significant associations with DED diagnosis and DED symptoms. The use of contact lenses, hypercholesterolaemia and stroke were only significantly associated with DED symptoms. When we attempted to examine risk factors in those participants under 50 years of age (DED prevalence 4.0% in 932 individuals), age, IBS and thyroid disease were the only factors that remained significantly associated, although the study is underpowered to find additional factors in this age group.

Table 2

Association between dry eye disease (DED) cases*, dry eye symptoms and putative risk factors

The subsample of 681 participants who also completed the OSDI and self-perceived health question (see table 1) had a mean age of 60.3 (SD 10.3) years. The prevalence of a DED diagnosis together with current use of artificial eye drops was 10.6% (OSDI total score 21.8 vs 7.4 in controls, p<0.0005). The prevalence of DED symptoms was 22.6% (OSDI total score 19.6 vs 6.7 in controls, p<0.0005). Subjects with DED symptoms scored 0.3 points lower on the 5-point scale of self-perceived health than controls (3.5 vs 3.8, p=0.001). This difference was present in subjects below 50 years of age (3.7 vs 4.0) and above 50 years of age (3.4 vs 3.7). For every point increase on the OSDI total score subjects scored a significant 0.014 points lower on self-perceived health scores (p<0.005). All 12 items on the OSDI questionnaire were associated with lower self-perceived health scores, including activities such as problems with watching TV (b=−0.28, p<0.0005), driving (b=-0.15, p<0.0005), computer work (b=−0.13, p=0.006) and ocular discomfort in places with low humidity (b=−0.23, p<0.0005) and with air conditioning (b=−0.21, p<0.0005).


The current study is the first population-based study of DED in the UK. The focus of this study was on women. It revealed an increasing prevalence of DED with increasing age, most notably from the sixth decade. Several known risk factors have been confirmed, and new risk factors have been discovered. DED symptoms are associated with lower self-perceived health.

We have found a strong significant association between DED and cataract surgery, showing a 70% higher prevalence of DED in subjects who had undergone cataract surgery. DED after ophthalmic surgery has been described, particularly after refractive surgery.12 Decreased corneal sensation and changes in tear physiology after phacoemulsification have been found,13 as well as a significantly decreased goblet cell density after cataract surgery.14 Age-related macular degeneration and glaucoma also showed a (near significant) increased risk of 30–70% of having DED. Glaucoma has been associated with DED before,8 as have glaucoma medications, particularly eye drops containing preservatives.15 Obviously there is a risk of ascertainment bias, as patients under the care of an ophthalmologist for other diseases likely have a higher chance of being diagnosed with DED and/or being prescribed artificial tears than patients who do not visit an ophthalmologist, simply because their eyes are evaluated. Therefore we also focused on DED symptoms in the past 3 months. Age-related macular degeneration and glaucoma were also associated with a (non-significant) increased risk of DED symptoms. The use of contact lenses, a well-known risk factor,2 ,4 was associated with DED symptoms in our cohort. Due to the cross-sectional design of this study, it is not a contradiction that no association was found between the current use of contact lenses and a clinical diagnosis of DED, since patients with DED are often advised to avoid wearing contact lenses.

Other previously described risk factors that are confirmed by our study are immune-mediated diseases such as rheumatoid arthritis, thyroid diseases and allergies. In addition to allergy, we found significantly strong associations with the other atopic disorders eczema and asthma. The strong association of immune-mediated and atopic disorders with DED diagnosis and symptoms might be the result of altered inflammatory mediators on the ocular surface.1 However, we cannot exclude that allergic eye diseases could be wrongly diagnosed as dry eye, or vice versa, by some practitioners. We found a significant association between fertility problems and the existence of a dry eye diagnosis, possibly reflecting the hormonal influence on DED. Several hormonal mechanisms have been proposed to explain this association, including oestrogens, progestagens, androgens and prolactin. For an extensive review on hormones and their relationship with dry eye, we refer to Sullivan.16 We also found a strong significant association with osteoarthritis, as have others.9 While the association of DED with rheumatoid arthritis can be linked to autoimmune processes, this is not possible with osteoarthritis. A possible explanation may lie in the fact that persons diagnosed with osteoarthritis show lower pain thresholds.17 Interestingly, we have found increased thermal pain sensitivity in patients with DED symptoms.10 Due to this increased pain sensitivity subjects may complain earlier of their DED and osteoarthritis than other persons with similar abnormalities. This association may point to a common aetiology between DED and pain.

Moreover, the strongest associations we found were with chronic pain syndromes: pelvic pain, IBS and CWPS (previously termed fibromyalgia). Patients with pelvic pain, IBS and CWPS have been associated with somatisation or psychosomatic symptoms.18 These patients show higher rates of several concomitant medical and psychiatric disorders, including migraine and depression, two conditions that were also highly associated with DED in this study. The various disorders have been hypothesised to share a common aetiology.18 The findings of this study suggest that DED might be added to this spectrum of disorders. Indeed, we recently explored this relationship in a multivariate twin study, showing evidence that a common pathway with shared genetic factors underlies DED and the chronic pain syndromes pelvic pain, IBS, CWPS.19 Interestingly, Türkyilmaz et al20 have shown an increased tear osmolarity and lowered tear break-up time and Schirmer's test value in patients with CWPS without any concomitant (rheumatological) disease, suggesting a possible pathophysiological link next to chronic pain disposition.

Depression has been associated with DED symptoms in several studies, but has never been associated with objective tests such as tear break-up time, corneal staining, Schirmer's test and meibomian gland dysfunction.21 ,22 Several mechanisms may underlie the association of DED with depression, such as an increase of depressive symptoms due to DED or an increase of perception of DED symptoms due to depression. In addition, depression and DED have common risk factors including the involvement of sex hormones,16 ,23 and the lack of ω-3 polyunsaturated fatty acids,24 ,25 which may lead to increased production of inflammatory cytokines. In addition, chronic depression itself can enhance acute and chronic proinflammatory cytokine production, which might worsen DED.21 The association of DED with migraine and depression may also be explained by its relation with pain syndromes such as IBS and CWPS. If we correct for IBS and CWPS, next to age, in a multivariable model, depression and migraine were no longer associated with a DED diagnosis (data not shown). In our cohort, subjects with migraine and depression show a two-times (significant) higher prevalence of IBS and CWPS.

A limitation of this study is that the diagnoses tested (dry eye and risk factors) are self-reported and not validated. As an example, dry eye may be confused by a clinician with allergy or (blepharo)conjunctivitis. However, the prevalence of DED in our cohort using questionnaire-based diagnostic criteria is similar to that found in epidemiological studies of women of similar age.2 ,4 While the results in our study are based on a twin cohort, random selection of one individual within each twin pair did not alter the effect size of the results (data not shown). An important additional limitation is that we did not have data on concurrent use of non-dry eye medication. Some of the associations we found may be induced or biased by the use of certain medications, such as antihistamines, antidiuretics and antidepressants, that have been associated with DED as well.1 ,2 Meibomian gland dysfunction is likely to have been underdiagnosed and investigated using our methodology of questionnaires in a population-based survey. Furthermore, the exclusion of participants who had positive responses to only one of two dry eye questions might introduce bias and artificially inflate the effect size of the risk factors. We therefore additionally performed a sensitivity analysis, allocating excluded participants as either (1) a case or (2) a control. In both alternative analyses similar ORs of the risk factors were found as in the original analysis. Two risk factors (fertility problems and diabetes) shifted across the statistical significance threshold; all other significant risk factors were the same as in the original analysis. Zygosity and ethnicity did not differ between those excluded persons and the cases and controls. So we have no reason to believe that the allocation of cases and controls introduces bias to the interpretation of our findings.

In conclusion, we have shown that DED is a very common problem among women in a population-representative cohort in the UK, becomes more prevalent with advancing age and has a significant effect on self-perceived health. Besides confirming some well-known risk factors, this study has found new associations of DED, raising the possibility that altered pain perception and psychological and somatisation factors influence DED and its symptomatology.


Supplementary materials


  • Contributors JV, PGH and CJH: writing the paper and data analysis; DK: data collection.

  • Funding TwinsUK receives support from the Wellcome Trust, EU and the National Institute for Health Research (NIHR) funded BioResource, Clinical Research Facility and Biomedical Research Centre based at Guy’s and St Thomas’ NHS Foundation Trust and King’s College London.

  • Competing interests PGH is a Fight for Sight Early Career Investigator Award holder.

  • Ethics approval Local Research Ethics Committee (07/h0802/84).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement TwinsUK resource access committee reviews all proposals for data access (