Customize your JAMA Network experience by selecting one or more topics from the list below.
Schaumberg DA, Buring JE, Sullivan DA, Dana MR. Hormone Replacement Therapy and Dry Eye Syndrome. JAMA. 2001;286(17):2114–2119. doi:10.1001/jama.286.17.2114
Author Affiliations: Departments of Medicine (Dr Schaumberg), Ambulatory Care and Prevention (Dr Buring), and Ophthalmology (Drs Sullivan and Dana), Division of Preventive Medicine, Brigham and Women's Hospital (Drs Schaumberg and Buring), and Schepens Eye Research Institute (Drs Schaumberg, Sullivan, and Dana), Harvard Medical School, and Department of Epidemiology, Harvard School of Public Health (Dr Buring), Boston, Mass.
Context Postmenopausal hormone replacement therapy (HRT) use is common in the
United States. Some research suggests that estrogen may have detrimental effects
on the tear film and could influence the development of dry eye syndrome,
but few data are available on this relationship.
Objective To determine the relationship of HRT and dry eye syndrome.
Design, Setting, and Participants The Women's Health Study, a large cohort study in which 25 665
postmenopausal women provided information about use of HRT at baseline (1992),
12, and 36 months and dry eye syndrome at 48 months.
Main Outcome Measures (1) Clinically diagnosed dry eye syndrome, as reported by participants;
(2) severe symptoms (both ocular dryness and irritation either constantly
or often); and (3) either clinically diagnosed dry eye syndrome or severe
symptoms, compared between women who used HRT vs those who did not.
Results For the combined end point of either clinically diagnosed dry eye syndrome
or severe symptoms, the multivariable-adjusted odds ratios were 1.69 (95%
confidence interval [CI], 1.49-1.91) for estrogen use alone and 1.29 (95%
CI, 1.13-1.48) for estrogen plus progesterone/progestin use compared with
no HRT use. Each 3-year increase in the duration of HRT use was associated
with a significant 15% (95% CI, 11%-19%) elevation in risk of clinically diagnosed
dry eye syndrome or severe symptoms. Results were similar for the combined
end point of clinically diagnosed dry eye syndrome and severe symptoms.
Conclusions These data suggest that women who use HRT, particularly estrogen alone,
are at increased risk of dry eye syndrome. Physicians caring for women who
are taking or considering HRT should be apprised of this potential complication.
Dry eye syndrome, or keratoconjunctivitis sicca, damages the ocular
surface and can cause debilitating symptoms of dryness and irritation, which
may result in psychological comorbidity and reduced work capacity.1-4 Dry eye
syndrome is associated with an enhanced risk of corneal infection, and, when
severe, can cause permanent visual impairment.2,3
Treatments for dry eye syndrome are generally costly and inadequate, and many
patients are unable to find satisfactory relief from their symptoms.5 Finally, dry eye syndrome accounts for a substantial
burden on the health care system,5 comprising
one of the leading causes of patient visits to both ophthalmologists and optometrists.2
Hormone replacement therapy (HRT) is used by an estimated 38% of postmenopausal
women in the United States.6 It has been shown
to have a clear role in the treatment of a variety of menopausal symptoms,7 and may confer other health benefits.8-11
However, some deleterious effects of HRT are increasingly recognized,12-14 and estrogen may
have adverse effects on the ocular surface.15-17
Despite this, virtually no data are available on the relationship of HRT and
dry eye syndrome. Therefore, we examined this relationship in the Women's
The Women's Health Study is a randomized trial among 39 876 health
professionals (aged 45 to 84 years in 1992) to assess the benefits and risks
of aspirin and vitamin E in the prevention of cardiovascular disease and cancer.18 Women were also initially randomized to beta carotene,
but this component of the trial was terminated after an average treatment
time of 22.8 months. To be eligible for the Women's Health Study, women must
have been postmenopausal or have no intention of becoming pregnant. At baseline,
all participants were free of cancer (except possibly nonmelanoma skin cancer),
myocardial infarction, stroke, transient cerebral ischemia, liver disease,
renal disease, peptic ulcer, or gout. Women using anticoagulants, corticosteroids,
or supplements of vitamins A, E, or beta carotene were also excluded. Participants
completed annual questionnaires reporting health-related exposures and any
health outcomes experienced over the previous year.
At baseline, participants reported demographic information including
age, race/ethnicity, educational level, and household income, as well as a
detailed medical history and information on lifestyle factors. Women reported
their HRT use at baseline and at 12 and 36 months of follow-up. We classified
postmenopausal women by their use at 36 months as either never or ever users
of HRT. We further classified ever users as using estrogen alone, or estrogen
combined with progesterone/progestins based on their most recent use pattern.
On the 4-year follow-up questionnaire we included 3 questions to assess
dry eye syndrome: How often do your eyes feel dry (not wet enough)? How often
do your eyes feel irritated? and Have you ever been diagnosed by a clinician
as having dry eye syndrome? The 2 questions pertaining to symptoms had possible
answers of constantly, often, sometimes, or never. These 2 questions alone
were previously found to have a sensitivity of 60% coupled with a specificity
of 94% compared with clinical diagnosis of dry eye syndrome, and to provide
nearly the same predictability as a 14-item questionnaire.19
We defined 3 outcome measures for dry eye syndrome. We defined clinically
diagnosed dry eye syndrome as a self-reported diagnosis of dry eye syndrome
by a clinician, and severe symptoms as a report of both dryness and irritation
either constantly or often. We also formed a composite end point of either
a previous clinical diagnosis or severe symptoms of dry eye syndrome.
We used χ2 tests to examine the relationship of HRT with
several potential determinants of its use as well as with dry eye syndrome.
We then constructed multivariable logistic regression models (separate models
for each definition of dry eye syndrome) to obtain odds ratios (ORs) and 95%
confidence intervals (CIs) to estimate the effects of estrogen use only, and
estrogen plus progesterone/progestin. We initially adjusted for age in 5-year
categories and, since subjects were participants in a randomized trial, randomized
treatment assignments (aspirin vs placebo, vitamin E vs placebo, and beta
carotene vs placebo). We then extended the models to account for predictors
of HRT, as well as other medical conditions that may have influenced the prevalence
of dry eye syndrome.
Although our primary analysis was restricted to postmenopausal women,
we also fit models in which premenopausal women formed the reference group.
In additional analyses, we fit models to examine whether the dose of estrogen
or progesterone/progestin or duration of HRT were related to dry eye syndrome.
There is evidence that androgens are protective against dry eye syndrome.20,21 Since women who had oophorectomy
would be expected to have lower androgen levels22
and to be more likely to take HRT, we also conducted separate analyses among
women based on oophorectomy history.
Finally, to address the issue of the timing of the onset of dry eye
syndrome relative to initiation of HRT, we conducted an additional analysis
for the end point of clinically diagnosed dry eye syndrome, in which we excluded
women who were diagnosed with dry eye syndrome prior to the initiation of
HRT. In this analysis, we chose an index date of 10 years prior to our assessment
of dry eye syndrome and excluded all women with a date of diagnosis prior
to this index date. Beginning from the index date, we then constructed a series
of consecutive 1-year intervals, which included data from all women who remained
free of dry eye syndrome at the beginning of the interval. We used data on
duration of HRT to estimate the time of initiation of therapy to determine
each participants' HRT status at the beginning of each interval, and data
on the date of diagnosis of dry eye syndrome to assign the diagnosis to the
interval in which it occurred. We used Cox proportional hazards models to
obtain estimates of the relative risk and 95% CI associated with HRT.
Information about dry eye syndrome was provided by 36 995 (93%)
of the 39 876 women enrolled in the Women's Health Study. Among the women
with data on dry eye syndrome, 25 665 (69%) were postmenopausal. We excluded
from further analyses 156 women who were taking either vaginal estrogen or
progesterone alone, as well as 120 women for whom data on HRT were unavailable.
Of the remaining 25 389 women, 61.1% had ever taken HRT and 90% of these
women were current users. As expected, HRT was related to a number of demographic
and social characteristics, being more common among younger women, women who
identified themselves as either white or Asian, and women with higher levels
of education and household income (Table
1). Use of HRT also varied by geographic region with the highest
prevalence in the West, and the lowest levels in the Northeast. Women who
had taken HRT were also more likely to have had an eye examination in the
past 2 years.
Use of HRT was significantly related to the prevalence of dry eye syndrome
(Figure 1). Considering the prevalence
of either clinically diagnosed dry eye syndrome or severe symptoms, women
who never used HRT had the lowest prevalence (5.9%). Women who used estrogen
alone had the highest prevalence (9.1%), and women who used a combination
of estrogen plus progesterone/progestin had a prevalence that was intermediate
between never users and users of estrogen alone (6.7%). Relationships were
similar for severe symptoms and clinically diagnosed dry eye syndrome.
After adjusting for age and randomized treatment assignments, HRT was
still significantly associated with clinically diagnosed dry eye syndrome
for estrogen alone (OR, 1.70 [95% CI, 1.49-1.95] and for estrogen and progesterone/progestin
OR, 1.30 [95% CI, 1.12-1.50]); severe symptoms (OR, 1.72 [95% CI, 1.46-2.03]
and OR, 1.24 [95% CI, 1.04-1.48]); and the combined end point of either clinically
diagnosed dry eye syndrome or severe symptoms (OR, 1.69 [95% CI, 1.51-1.90]
and OR, 1.27 [95% CI, 1.12-1.44]), respectively. Further adjustment for race/ethnicity,
geographic region, educational level, household income, and frequency of eye
examinations had little impact on these findings. Additional adjustment for
diabetes mellitus, hypertension, rheumatoid arthritis, and other connective
tissue diseases also had no effect (Table
2). Compared with no HRT use, the multivariable-adjusted ORs for
the combined end point of clinically diagnosed dry eye syndrome or severe
symptoms were 1.69 (95% CI, 1.49-1.91) for estrogen use alone and 1.29 (95%
CI, 1.13-1.48) for estrogen plus progesterone/progestin use.
In models that included premenopausal women as the reference group,
the multivariable-adjusted ORs for clinically diagnosed dry eye syndrome or
severe symptoms were 1.02 (95% CI, 0.86-1.22) for postmenopausal women who
never used HRT; 1.71 (95% CI, 1.46-2.00) for postmenopausal women who used
estrogen alone; and 1.29 (95% CI, 1.10-1.51) for postmenopausal women who
used estrogen plus progesterone/progestin.
In models examining dose, the risk of dry eye syndrome was elevated
compared with nonusers in all women who took estrogen, including women who
used less than 1 mg/d, the lowest prescribed doses (multivariable-adjusted
OR, 1.73 [95% CI, 1.25-2.41]), and there was no clear dose-response relationship.
Use of progesterone/progestin in combination with estrogen resulted in lower
risks of dry eye syndrome compared with those associated with use of estrogen
alone regardless of progesterone dose. For example, in women taking the lowest
doses of progesterone/progestin in combination with the lowest doses of estrogen,
the multivariable-adjusted OR was reduced from 1.73 for estrogen alone to
1.31 (95% CI, 0.80-2.14) for the combination. Duration of HRT was significantly
associated with a multivariable-adjusted 15% higher risk (95% CI, 11%-19%)
of dry eye syndrome for each 3-year increase in the duration of HRT use.
Among the subgroup of postmenopausal women without a history of oophorectomy,
the multivariable-adjusted ORs for the end point of clinically diagnosed dry
eye syndrome or severe symptoms were 1.38 (95% CI, 1.04-1.83) for estrogen
alone and 1.32 (95% CI, 1.14-1.54) for estrogen plus progesterone/progestin.
Among postmenopausal women with a history of oophorectomy, the multivariable-adjusted
ORs were 1.53 (95% CI, 1.24-1.89) for estrogen alone and 1.24 (95% CI, 0.79-1.94)
for estrogen plus progesterone/progestin.
Finally, in models examining the timing of the initiation of HRT relative
to the diagnosis of dry eye syndrome, we observed a higher incidence of clinically
diagnosed dry eye syndrome among women who were free of clinically diagnosed
dry eye syndrome at the time they began using HRT (multivariable-adjusted
relative risk, 1.48 [95% CI, 1.27-1.72] for estrogen alone; relative risk,
1.15 [95% CI, 0.97-1.37] for estrogen plus progesterone/progestin).
In this study, postmenopausal women who used HRT had higher prevalences
of dry eye syndrome than never users (estrogen alone, 69%; estrogen plus progesterone/progestin,
29%). Adjusting for age and other factors, postmenopausal women who had never
used HRT did not differ from premenopausal women in the prevalence of dry
eye syndrome. The relationship of HRT and dry eye syndrome was consistent
for all definitions of dry eye syndrome used in the present study, and held
for clinically diagnosed cases diagnosed after the initiation of therapy.
The longer the duration of HRT, the higher the risk of dry eye syndrome.
Since we were not able to determine if initiation of HRT preceded the
onset of dry eye syndrome, the relationships we observed may reflect a higher
tendency of women with dry eye syndrome to be prescribed HRT. However, given
the lack of evidence of any beneficial effect of replacement hormones in this
context, its prescription specifically for dry eye syndrome is not likely
to be common and certainly cannot be considered a standard of practice. Moreover,
when we excluded cases that were clinically diagnosed prior to the initiation
of HRT, we continued to observe a significantly elevated risk of subsequent
dry eye syndrome among women who took exogenous estrogen. Although we could
not address the issue of timing for our analysis based on dry eye symptoms,
the similarity of our findings for symptoms compared with clinically diagnosed
cases as well as the consistently stronger relationship with estrogen alone
and significant increased risk with longer duration HRT use argue against
this as a major source of bias. These factors, and the fact that participants
were unaware of our hypothesis when they provided information about HRT use
and dry eye syndrome (collected at different times and on different questionnaires),
also argue against the possibility that women taking HRT were more likely
to think that such therapy caused dry eye syndrome and therefore were more
likely to report dry eye symptoms.
Residual confounding is a concern in any epidemiological study. In the
present study, we were not able to control for factors such as contact lens
use or use of other medications that might lead to an increased frequency
of dry eye syndrome. However, given the high prevalence of HRT and the magnitude
of the observed effects, any extraneous factor would need to be prevalent
as well as strongly related to both HRT and dry eye syndrome to explain the
observed associations. With regard to contact lens use, in a subgroup of 393
women, we determined that HRT was not associated with contact lens use (26.8%
of never users wore contact lenses vs 24.8% of HRT users), making residual
confounding by this factor unlikely. Moreover, control for medical conditions
such as hypertension, diabetes, rheumatoid arthritis, and other connective
tissue diseases had little impact on our findings.
An additional consideration relates to our use of a questionnaire-based
assessment of dry eye syndrome, although there is consensus among both researchers2 and clinicians23,24
that ascertainment of dry eye symptoms provides important information. In
fact, assessment of symptoms was determined to be the single most important
test for dry eye syndrome identified by clinicians in practice.23,24
An expert panel also identified these symptoms to be the sine qua non of dry
eye syndrome.2 This seems appropriate since
ocular surface damage rarely reaches clinical importance in the absence of
symptoms,25 and a major goal of therapy for
dry eye syndrome is the relief of debilitating symptoms. Moss et al26 identified expected relationships when using self-reported
dry eye syndrome in epidemiological studies. In the present study, we used
a validated questionnaire to assess symptoms of dry eye syndrome, and strict
criteria to identify women as having dry eye syndrome based on symptoms alone.
We also assessed previous clinical diagnoses of dry eye syndrome, which should
have helped us identify participants with treated dry eye syndrome who had
received some relief from their symptoms, as well as cases with only milder
symptoms. Because we were not able to examine study participants, however,
estimates could have been biased by a higher likelihood of diagnosis among
women using HRT, although controlling for more frequent eye examinations did
not have any impact on our findings. This explanation also seems unlikely
given the significant relationship of HRT with symptoms alone, the stronger
effect of estrogen taken alone, and the significant increase in risk with
longer duration of HRT use.
Despite the common occurrence of dry eye syndrome, basic epidemiological
data are limited. Clinical observations suggest, and most epidemiological
that dry eye syndrome is more common in women, a finding that would be consistent
with either a detrimental effect of estrogen or a beneficial role of androgens,1,21 or both. Indeed, it may be the balance
of androgens and estrogen that is important in determining risk of dry eye
syndrome. Since women with oophorectomy would be expected to have lower androgen
levels22 and are also more likely to be prescribed
estrogen replacement, we were concerned that low androgen levels might have
confounded the relationship of exogenous estrogen with dry eye syndrome. However,
when we looked separately among the subgroups of women based on history of
oophorectomy, we observed elevated risks of dry eye syndrome associated with
estrogen in each subgroup, suggesting that these relationships were not likely
to be purely a consequence of confounding by low androgen levels.
There are few epidemiological studies that directly assess the potential
relationship of exogenous estrogen use with dry eye syndrome, and none that
has examined the relationship in as much detail as the present study. Two
studies26,27 reported that there
was no statistically significant relationship of HRT with the presence of
self-reported dry eye symptoms. However, the data were not actually presented
in either study, estrogen and estrogen plus progesterone/progestin were not
examined separately, and it is unlikely that either study had sufficient statistical
power to detect an association of the magnitude we observed.
Strengths of the present study include its large sample size and the
prevalent use of HRT, which provided a high degree of precision for our estimates
of an association with dry eye syndrome. We also used a validated questionnaire-based
assessment of dry eye symptoms with high specificity for identifying subjects
with dry eye syndrome. In addition, we obtained information on clinical diagnoses
of dry eye syndrome from our population of knowledgeable female health professionals,
and reporting of medical diagnoses has proven reliable among such populations.29,30 Information on HRT was obtained without
knowledge of dry eye status (and vice-versa). The results of the present study
consistently showed a higher prevalence of dry eye syndrome among women who
used HRT, regardless of the way in which we defined dry eye syndrome. As would
be expected if the relationship were real rather than spurious, similar findings
were observed for clinically diagnosed dry eye syndrome, severe symptoms,
or either condition. Moreover, there was a significant trend of increasing
prevalence of dry eye syndrome with longer duration of HRT use.
Basic research suggests that sex hormone levels may influence both the
lacrimal and meibomian glands.15,21
Laboratory and preliminary clinical studies suggest that whereas androgens
have a beneficial influence on lacrimal and meibomian gland function,15,21 estrogen may play a role in exacerbating
dry eye syndrome.15-17,21,31,32
Given our findings as well as the known inhibitory effects of estrogen on
other sebaceous glands,33 further study of
the effects of estrogen on the function of the meibomian gland—a large
sebaceous gland containing estrogen receptors34—would
be interesting. Moreover, the apparently beneficial modifying effect of progesterone/progestin
on the relationship of estrogen with dry eye syndrome requires further study.
In summary, the present study suggests that postmenopausal women who
use HRT have a higher prevalence of dry eye syndrome compared with those who
have never used HRT, and this is particularly true of women who used estrogen
alone. Given these findings and the high prevalence of HRT in the United States,
further studies of the effects of sex-steroid hormones on dry eye syndrome
are recommended. Meanwhile, physicians caring for women who are taking or
are considering HRT should be informed of the potential increased risk of
dry eye syndrome with this therapy.