Dry eye has been called the back pain of eye care, one of those nebulous conditions of many, often unknown, etiologies that is a source of great frustration to patients and their eye care providers. Though rarely affecting vision, it can have an adverse effect on quality of life.1–4 It is common in elderly populations.5–10 Although it is often associated with specific factors such as rheumatoid arthritis, other autoimmune diseases, and certain medications, there are few large population-based epidemiologic studies of dry eye that have examined these associations.5–10 Most such studies are of the prevalence of dry eye.5–10 Being cross-sectional in design, there is no information of the antecedent-consequent association between the risk factors and the condition. For this, incidence studies are needed.
Earlier, we reported on the 5-year incidence of dry eye in the Beaver Dam Eye Study.11 With this paper, we extend those previous results to 10 years of follow-up.
The methods for identifying the Beaver Dam Eye Study population, reasons for non-participation, and comparisons between participants and non-participants have previously been published.12,13 Briefly, 5924 residents between the ages of 43 and 86 years were identified in a private census of Beaver Dam, WI, conducted from September 15, 1987, to May 4, 1988. During the period of 30 months from March 1, 1988, 4926 (83.1%) of the eligible residents were examined.12 Beginning on March 1, 1993, 5-year follow-up examinations began during which 3684 of 4368 surviving baseline subjects were examined. In addition, 38 eligible residents who had not participated in the baseline examination were examined. It was at this examination that information on the presence of dry eye was first collected. Thus, this examination is considered to be the baseline for the purposes of the present analysis. Fig. 1 presents a schematic of further study participation of the 3722 persons examined in 1993 to 1995. Of these, 534 were not at risk for incidence of dry eye because they reported already having it, 3169 were at risk, and 19 were missing information about dry eye. Subsequently, 2430 of the 3169 persons at risk were examined or interviewed in 1998 to 2000 of whom 322 reported the presence of dry eye, 2092 did not, and 16 provided no information. In 2003 to 2005, 1679 of the 2092 persons still at risk were examined or interviewed of whom 160 reported the presence of dry eye, 1516 did not, and 3 provided no information. At all examinations, substantial attempts were made to examine participants in the same order so as to preserve an average five year interval. Comparisons of subjects who participated in subsequent examinations, who did not participate, and who died have been presented previously.11 Briefly, those who did not participate were older and had higher systolic blood pressure than participants. Otherwise the two groups were similar with respect to gender, diastolic blood pressure, body mass, total and high density lipoprotein (HDL) cholesterol, and history of cardiovascular disease and diabetes. Subjects who died before subsequent examinations were older, were more likely to be male and have a history of cardiovascular disease or diabetes, had higher systolic blood pressure, and had lower diastolic blood pressure, body mass, and total and HDL cholesterol.
All examinations have followed similar protocols, which were approved by the Institutional Review Board at the University of Wisconsin-Madison in conformity with all federal and state laws, and the study was in adherence to the tenets of the Declaration of Helsinki. Informed signed consent was obtained for each participant before each examination. The examination included a medical history questionnaire; measurement of height, weight, and blood pressure; and collection of urine and blood samples for a series of standard laboratory tests.
Systolic and diastolic blood pressures were the average of two measurements. Hypertension was defined as a systolic blood pressure of 160 mm Hg or greater, a diastolic blood pressure of 95 mm Hg or greater, or a history of hypertension with use of antihypertension medications. Body mass was defined as weight (kg) divided by the square of height (m). A subject was considered to have diabetes if he or she gave a history of diabetes mellitus treated with insulin, oral hypoglycemic agents, or diet, or was newly diagnosed during participation in the study. The criterion for diagnosis was a glycosylated hemoglobin value greater than two standard deviations above the mean for a given age-sex group and a random blood sugar of >11.1 mmol/l. Arthritis, fractures, osteoporosis, gout, and thyroid disorder were determined by history. Participants were asked to bring to the examination all medications, both prescription and over-the-counter, that they were taking regularly. The examiner asked if there were other medications being taken but were not brought. If there were, the participant was asked to call the examiner with the medication name. When necessary, participants, their physicians, and their pharmacies were called to verify medication use. Each active ingredient of a drug was assigned a code reflecting that of the American Hospital Formulary Service.14 Current or past consumption of four or more servings of alcoholic beverages daily constituted heavy drinking. Average weekly alcohol consumption was computed as the grams of alcohol from each 0.355 l (12 oz) serving of beer, 0.118 l (4 oz) serving of wine, and 0.044 l (1.5 oz) serving of liquor or distilled spirits. Each serving of beer, wine, and liquor was assigned an alcohol content of 12.96 g, 11.48 g, and 14.00 g of alcohol, respectively. An individual who had smoked at least 100 cigarettes in his life was considered to be a current or ex-smoker. Pack-years smoked were computed as the number of packs of 20 cigarettes smoked each day times the number of years smoked. The average daily consumption of caffeine was computed as the sum of milligrams of caffeine from each 0.237 l (8 oz) serving of brewed coffee (103 mg), instant coffee (57 mg), hot or iced tea (36 mg), hot chocolate (6 mg), and caffeine containing soda (46 mg).
History of dry eye was not determined at the initial Beaver Dam Eye Study examination in 1988 to 1990. At the 5-, 10-, and 15-year follow-up examinations, the presence of dry eye was determined by subject self-report. Dry eye was defined as a positive response to the question, “For the past 3 months or longer, have you had dry eyes? ” This was described as a “foreign body sensation with itching and burning, sandy feeling, not related to allergy” for subjects needing further prompting. Because history of dry eye was not obtained until the 1993 to 1995 examination, that juncture is regarded as the “baseline” for these analyses. Except as noted below, all covariate information pertains to that point. To be included in the analysis, an individual must have participated in the 1993 to 1995 examination, reported not having dry eye at that time, participated in the 1998 to 2000 examination, and responded to the dry eye question at that time. Incidence of dry eye was defined as a positive report of dry eye at the 10- or 15-year examination or interview in persons who did not report having dry eye at the 5-year examination. Some participants who were still at risk for dry eye at the 10-year follow-up examination did not return for the 15-year follow-up owing to refusal to participate or death. These are censored observations. To calculate the 10-year cumulative incidence of dry eye while including these censored observations, the product-limit method was used.15 Tests of differences in rates of incidence or trends in incidence were performed with the Mantel-Haenszel procedure stratified by observation interval.16 Age-adjusted incidence rates were computed by the direct method.17 Proportions of 0.4258, 0.3057, 0.2084, and 0.0601 for the age groups 48 to 59, 60 to 69, 70 to 79, and 80 to 91, respectively, were used for the standard population. These proportions reflect the age distribution of the set of participants at risk for incidence of dry eye. Multivariable analyses of factors associated with incidence of dry eye were performed with discrete linear logistic regression.18 These age-adjusted and multivariable analyses used risk factor information only from the 5-year baseline examination. Additional models using time-dependent covariates were analyzed in the following manner. For the time period between the 5- and 10-year examinations, values of risk factors at the 5-year baseline examination were used. For the period between the 10- and 15-year examinations for persons still at risk, values of risk factors at the 10-year examination were used. This procedure has the advantage of using the maximum risk factor information available because it accounts for changes in risk factors that occur during follow-up.19
Of the 2827 subjects who participated in the 5- and 10-year examinations, 388 had reported a history of dry eye at baseline, and 25 were missing information on dry eye at baseline or subsequent examinations. Excluding these subjects left 2414 persons at risk for incidence of dry eye. Age among subjects in this group ranged from 48 to 91 years with a mean ± SD of 63 ± 10 years. Men composed 44% of the cohort, and 99% were white.
Dry eye was reported in 322 of the 2414 at risk subjects in the first five years. An additional 160 cases of dry eye were reported in 1676 at risk subjects who completed the second five year period. This resulted in a cumulative 10-year incidence of 21.6% (95% confidence interval, 19.9 to 23.3%). Incidence increased from 17.3% in subjects 48 to 59 years of age to 28.0% in those 80 years or older (p < 0.001) (Fig. 2). Incidence was also greater in women (25.0%) compared to men (17.2%, p < 0.001). This result remained after adjusting for age (p < 0.001) (Fig. 2).
Because many of the potential risk factors examined are associated with age as is the incidence of dry eye, further analyses were adjusted by age. Table 1 presents the age-adjusted cumulative 10-year incidence of dry eye by selected subject characteristics. Persons with a history of allergy or thyroid disease had higher incidence of dry eye. In the case of thyroid disease, this higher incidence was confined to those who were not being treated with thyroid hormone (34.8%). This group included three subjects who were receiving medication for hyperthyroidism. Those being treated with thyroid hormone had an age-adjusted incidence (21.8%) similar to those without thyroid disease (21.0%). Persons with a history of arthritis or possible arthritis had higher incidence of dry eye. In addition, we found that people with dry eye at baseline had higher incidence of arthritis over the next 10 years than people without (49.5% vs. 39.4%, p = 0.02). Also, people with incidence of dry eye had higher incidence of arthritis than those without (45.5% vs. 38.1%, p = 0.03). Persons with diabetes had a slightly elevated but statistically non-significant incidence of dry eye. Other conditions which were not related to incidence of dry eye included history of gout, osteoporosis, and fractures (data not shown). Persons with higher serum total cholesterol tended to have a higher incidence of dry eye, but the trend did not reach statistical significance. No associations with incidence were found for serum HDL cholesterol, ratio of total to HDL cholesterol, body mass, systolic or diastolic blood pressure, hypertension, leukocyte count, and hematocrit level (data not shown).
The use of several medications was also associated with higher age-adjusted incidence of dry eye (Table 1). These included antihistamines, antianxiety medications, antidepressants, and oral steroids. Diuretics were also associated with higher incidence, but this difference did not quite reach statistical significance. Drugs which were not related to incidence included calcium channel blockers, cholesterol lowering medications, and hormone replacement medication in women (Table 1). Finally, vitamin users, either current or past, reported higher age-adjusted incidence of dry eye.
Among lifestyle factors, only alcohol consumption was associated with age-adjusted incidence of dry eye with lower incidence at higher consumption levels (Table 1). In addition, persons reporting themselves to be in excellent or good health compared to others their age had lower dry eye incidence than those who rated themselves to be in fair or poor health. Smoking history and sedentary lifestyle were not associated with dry eye incidence. Other lifestyle factors not associated with dry eye incidence included history of heavy drinking and caffeine use (Table 1).
Discrete linear logistic regression models were developed to determine which factors were independently associated with dry eye incidence. Age was included in every model. Other variables were selected in stepwise fashion. The results of this analysis are presented in Table 2. In addition to older age, subjects with poorer self-rated health, women, thyroid disease untreated with thyroid hormone, and using oral steroids were at higher risk for incident dry eye. Those with a sedentary lifestyle were at lower risk. The results of an additional analysis using time-dependent covariates are presented in Table 3. The same variables from the previous step were also significant in this model except that thyroid disease untreated with thyroid hormone was of borderline significance. In addition, persons using antidepressants were at higher risk for incidence of dry eye while those using angiotensin-converting enzyme (ACE) inhibitors were at lower risk. Also, persons using diuretics were at borderline higher risk.
To our knowledge, this paper and our earlier paper on 5-year incidence11 are the only studies of dry eye incidence in large population-based cohorts. The few other large observational studies have been cross-sectional or prevalence studies.5–10 Incidence studies, such as that reported herein, have the advantage that the potential risk factors examined were observed before the occurrence of dry eye. Thus, any risk factors found to be associated with later incidence of dry eye are more (though not necessarily) likely to be in the causal pathway as opposed to merely correlated with the outcome.
As a consequence of the lack of incidence studies, there are none to compare directly with ours. However, comparisons with prevalence studies are still useful in identifying important risk factors across studies. For instance, some studies report an increase in dry eye prevalence with age.6,8,20 Others do not.5,9,10 Our incidence results agree with the former studies. The lack of a trend in some studies may be due to a narrower age range.5,10 Women are often reported to have a higher prevalence of dry eye,6,7,9 but some studies find otherwise.5,8 Prospectively, we find higher dry eye incidence in women. The higher occurrence of dry eye in women may be due to hormonal influences.21,22 Some studies have suggested a detrimental effect of hormone replacement therapy.9,23,24 However, we found no association of dry eye incidence with a history of hormone replacement therapy.
Several studies have noted strong associations between dry eye and arthritis.6,9,25 Although we also found such an association in our earlier prevalence study,7 we failed to find an association between arthritis and dry eye incidence in our earlier paper.11 In the current analysis, we found 10-year age-adjusted incidence of dry eye to be associated with arthritis. The added statistical power associated with the additional 5 years of follow-up may have tipped the balance in favor of finding the association. However, it failed to reach statistical significance when controlling for other factors. We speculated in the earlier incidence paper that dry eye may develop before or currently with arthritis with the result of arthritis being associated with prevalence but not incidence of dry eye. We found this to be the case then,11 and it continues to be so in the current analysis. An alternative explanation for the absence of a relationship between arthritis and dry eye incidence is that we are not able to determine the type of arthritis present. Thus, any relationship between rheumatoid arthritis and dry eye incidence would be attenuated in our cohort.
Unlike the results of the 5-year incidence analysis,11 we found an association of thyroid disease with 10-year incidence of dry eye which is consistent with the prevalence results.7 Again, the power gained from the additional follow-up could explain this difference. Unlike in the earlier analysis, we also explored the effect of thyroid hormone treatment, finding that the association was restricted to subjects who were not being treated for their thyroid disease. Allowing that all but three subjects who were being treated for thyroid disease were receiving thyroid hormone presumably for hypothyroidism, it can be assumed that the group not receiving thyroid hormone also suffers predominately from hypothyroidism. Reduced tear production in hypothyroidism has been noted in one animal model.26 Other research has suggested an autoimmune disorder as a common link between thyroid disease and Sjögren syndrome, a major source of dry eye.27,28
An additional difference between the 5- and 10-year incidence results is in the relationship of diabetes to dry eye incidence. In the earlier paper, we found persons with diabetes to have a significantly higher age-adjusted incidence of dry eye. Although at the 10-year follow-up the incidence of dry eye was still higher in those with diabetes, the difference was no longer statistically significant. This may be due to random variation in the incidence rates. It may also be that the earlier result was a chance finding. Further research is needed.
Several classes of drugs, including diuretics, antihistamines, and psychotropics, are believed to be associated with prevalence of dry eye presumably owing to their drying properties.29–31 In the age-adjusted analysis, we found subjects using antidepressants, antianxiety drugs, antihistamines, and oral steroids to have a greater risk of dry eye incidence. Diuretics were of borderline significance. In the multivariable analysis, only oral steroids remained statistically significant. However, in the multivariable analysis with time-dependent covariates, antidepressants and diuretics in addition to oral steroids were significant or suggestive. This suggests that these medications have more effect in the short term. For instance, incidence of dry eye at 10 years may be too far removed from baseline to be influenced by medication use at that time or medication use may have changed. By also incorporating medication use at the 5-year follow-up, time-dependent covariate analysis brings the outcome closer to the antecedent risk factor.
Similarly, ACE inhibitors were not statistically significant except in the multivariable analysis with time-dependent covariates, differing from the 5-year results. However, this again suggests a short-term effect which, in this instance, was protective. This may be due to the reported anti-inflammatory effects of ACE inhibitors.32,33
Subjects leading a sedentary lifestyle were found to have a lower risk of dry eye incidence, or conversely, active persons were at higher risk. We know of no other reports of this association. Active people may be more exposed to environmental irritants such as sunlight, wind, and dust. Alternatively, this may be a chance finding.
This point leads to limitations of the study. First, because a number of associations have been examined, some statistically significant results may be due to chance. As suggested above, this may also be the source of differences between the 5- and 10-year results. Thus, it is important to consider the results in a context that includes other available evidence and biological credibility. Next, we relied on self-reports of the study participants to determine the presence of dry eye. No objective testing was performed. However, the tests commonly performed have been shown to be lacking in sensitivity and specificity.6,34 Thus, in the case of conditions such as dry eye, we believe the subjects themselves are best able to judge the presence of the condition. Furthermore, a recent report has established the value of short questionnaires for determining the presence of dry eye especially in the context of large epidemiologic studies where objective testing may not be practical.35 However, as a consequence of the absence of objective testing, we are not able to differentiate between tear-deficient and evaporative dry eye. Risk factors may be different for these two conditions. Thus, any associations present may be diluted in our analysis. In addition, differences between this study and others in how dry eye is determined make it difficult to contrast findings among them. However, this would apply to quantitative comparisons more than comparisons of general risk factor associations. Because our cohort represents an older population, it has experienced attrition due to death before and after the occurrence of dry eye. If the risk factors and dry eye are both associated with mortality, a bias in the results is possible. However, as we have shown previously that dry eye is not associated with subsequent mortality,11 we believe this is unlikely. Finally, as the cohort is a white, middle-class population, the results of this analysis may not be applicable to other populations differing in ethnicity, socioeconomic class, or geography. However, as it is population-based, our research is more likely to be representative of the general population than a clinic-based study.
In conclusion, incidence of dry eye is common in an older population, occurring at a cumulative rate of 21.6% after 10 years in persons aged 48 to 91 years. Few independently associated risk factors were found. Fewer still are subject to intervention. Several drug classes, including diuretics, antidepressants, and oral steroids, were associated with increased dry eye risk, whereas ACE inhibitors were associated with decreased risk. This suggests further study to examine the use of these agents as a way of relieving dry eye symptoms. Also, the finding of increased risk in persons with untreated hypothyroidism is interesting and may benefit from further research.
This study was supported by National Eye Institute, National Institute of Health, grant # 06594 (R. Klein and B.E.K. Klein) and, in part, by the Research to Prevent Blindness (Klein B.E.K., Klein R., Senior Scientific Investigator Awards), New York, NY.
Scot E. Moss
Department of Ophthalmology and Visual Sciences
University of Wisconsin Madison
610 N. Walnut Street, 4th Floor WARF
Madison, WI 53726
1. Thomas E, Hay EM, Hajeer A, Silman AJ. Sjogren's syndrome: a community-based study of prevalence and impact. Br J Rheumatol 1998;37:1069–76.
2. Nelson JD, Helms H, Fiscella R, Southwell Y, Hirsch JD. A new look at dry eye disease and its treatment. Adv Ther 2000;17:84–93.
3. Schiffman RM, Walt JG, Jacobsen G, Doyle JJ, Lebovics G, Sumner W. Utility assessment among patients with dry eye disease. Ophthalmology 2003;110:1412–9.
4. Mertzanis P, Abetz L, Rajagopalan K, Espindle D, Chalmers R, Snyder C, Caffery B, Edrington T, Simpson T, Nelson JD, Begley C. The relative burden of dry eye in patients’ lives: comparisons to a U.S. normative sample. Invest Ophthalmol Vis Sci 2005;46:46–50.
5. Schein OD, Munoz B, Tielsch JM, Bandeen-Roche K, West S. Prevalence of dry eye among the elderly. Am J Ophthalmol 1997;124:723–8.
6. McCarty CA, Bansal AK, Livingston PM, Stanislavsky YL, Taylor HR. The epidemiology of dry eye in Melbourne, Australia. Ophthalmology 1998;105:1114–9.
7. Moss SE, Klein R, Klein BE. Prevalence of and risk factors for dry eye syndrome. Arch Ophthalmol 2000;118:1264–8.
8. Lee AJ, Lee J, Saw SM, Gazzard G, Koh D, Widjaja D, Tan DT. Prevalence and risk factors associated with dry eye symptoms: a population based study in Indonesia. Br J Ophthalmol 2002;86:1347–51.
9. Chia EM, Mitchell P, Rochtchina E, Lee AJ, Maroun R, Wang JJ. Prevalence and associations of dry eye syndrome in an older population: the Blue Mountains Eye Study. Clin Experiment Ophthalmol 2003;31:229–32.
10. Lin PY, Cheng CY, Hsu WM, Tsai SY, Lin MW, Liu JH, Chou P. Association between symptoms and signs of dry eye among an elderly Chinese population in Taiwan: the Shihpai Eye Study. Invest Ophthalmol Vis Sci 2005;46:1593–8.
11. Moss SE, Klein R, Klein BE. Incidence of dry eye in an older population. Arch Ophthalmol 2004;122:369–73.
12. Klein R, Klein BE, Linton KL, De Mets DL. The Beaver Dam Eye Study: visual acuity. Ophthalmology 1991;98:1310–5.
13. Klein R, Klein BE, Lee KE. Changes in visual acuity in a population. The Beaver Dam Eye Study. Ophthalmology 1996;103:1169–78.
14. McEvoy GK. American Hospital Formulary Service: Drug Information 2000. Bethesda, MD: American Society of Hospital Pharmacists; 2000.
15. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–81.
16. Mantel N. Chi-square tests with one degree of freedom: extensions of the Mantel-Haenszel procedure. J Am Stat Assoc 1963;58:690–700.
17. Fleiss JL. Statistical Methods for Rates and Proportions. New York: Wiley; 1973.
18. Hosmer DW Jr, Lemeshow S. Applied Logistic Regression. New York: Wiley; 1989.
19. Selker HP, Raitt MH, Schmid CH, Laks MM, Beshansky JR, Griffith JL, Califf RM, Selvester RH, Maynard C, D'Agostino RB, Weaver WD. Time-dependent predictors of primary cardiac arrest in patients with acute myocardial infarction. Am J Cardiol 2003;91:280–6.
20. Schaumberg DA, Sullivan DA, Buring JE, Dana MR. Prevalence of dry eye syndrome among US women. Am J Ophthalmol 2003;136:318–26.
21. Mathers WD, Stovall D, Lane JA, Zimmerman MB, Johnson S. Menopause and tear function: the influence of prolactin and sex hormones on human tear production. Cornea 1998;17:353–8.
22. Smith JA, Vitale S, Reed GF, Grieshaber SA, Goodman LA, Vanderhoof VH, Calis KA, Nelson LM. Dry eye signs and symptoms in women with premature ovarian failure. Arch Ophthalmol 2004;122:151–6.
23. Schaumberg DA, Buring JE, Sullivan DA, Dana MR. Hormone replacement therapy and dry eye syndrome. JAMA 2001;286:2114–9.
24. Erdem U, Ozdegirmenci O, Sobaci E, Sobaci G, Goktolga U, Dagli S. Dry eye in post-menopausal women using hormone replacement therapy. Maturitas 2007;56:257–62.
25. Punjabi OS, Adyanthaya RS, Mhatre AD, Jehangir RP. Rheumatoid arthritis is a risk factor for dry eye in the Indian population. Ophthalmic Epidemiol 2006;13:379–84.
26. Williams DL, Pierce V, Mellor P, Heath MF. Reduced tear production in three canine endocrinopathies. J Small Anim Pract 2007;48:252–6.
27. Perez B, Kraus A, Lopez G, Cifuentes M, Alarcon-Segovia D. Autoimmune thyroid disease in primary Sjogren's syndrome. Am J Med 1995;99:480–4.
28. D'Arbonneau F, Ansart S, Le Berre R, Dueymes M, Youinou P, Pennec YL. Thyroid dysfunction in primary Sjogren's syndrome: a long-term followup study. Arthritis Rheum 2003;49:804–9.
29. Bergmann MT, Newman BL, Johnson NC Jr. The effect of a diuretic (hydrochlorothiazide) on tear production in humans. Am J Ophthalmol 1985;99:473–5.
30. Norn M. The effects of drugs on tear flow. Trans Ophthalmol Soc U K 1985;104(Pt 4):410–4.
31. Schein OD, Hochberg MC, Munoz B, Tielsch JM, Bandeen-Roche K, Provost T, Anhalt GJ, West S. Dry eye and dry mouth in the elderly: a population-based assessment. Arch Intern Med 1999;159:1359–63.
32. Peeters AC, Netea MG, Kullberg BJ, Thien T, van der Meer JW. The effect of renin-angiotensin system inhibitors on pro- and anti-inflammatory cytokine production. Immunology 1998;94:376–9.
33. Sheth T, Parker T, Block A, Hall C, Adam A, Pfeffer MA, Stewart DJ, Qian C, Rouleau JL. Comparison of the effects of omapatrilat and lisinopril on circulating neurohormones and cytokines in patients with chronic heart failure. Am J Cardiol 2002;90:496–500.
34. Schein OD, Tielsch JM, Munoz B, Bandeen-Roche K, West S. Relation between signs and symptoms of dry eye in the elderly. A population- based perspective. Ophthalmology 1997;104:1395–401.
35. Gulati A, Sullivan R, Buring JE, Sullivan DA, Dana R, Schaumberg DA. Validation and repeatability of a short questionnaire for dry eye syndrome. Am J Ophthalmol 2006;142:125–31.
Keywords:© 2008 American Academy of Optometry
dry eye; incidence; risk factors; epidemiology; population-based