Garbe E, Suissa S, LeLorier J. Association of Inhaled Corticosteroid Use With Cataract Extraction in Elderly Patients. JAMA. 1998;280(6):539-543. doi:10.1001/jama.280.6.539
From the Potsdam Institute of Pharmacoepidemiology and Technology Assessment, Potsdam, Germany (Dr Garbe); the Division of Clinical Epidemiology, Royal Victoria Hospital and the Department of Epidemiology and Biostatistics, McGill University (Dr Suissa) and the Centre de Recherche, Hôtel-Dieu de Montreal, Université de Montreal (Dr LeLorier), Montreal, Quebec.
Context.— The use of systemic corticosteroids is a known risk factor for the development
Objective.— To determine whether treatment with inhaled corticosteroids is associated
with cataract extraction in the elderly.
Design.— Case-control study.
Setting.— Quebec universal health insurance program for all elderly (provincial
health insurance plan database [RAMQ database]).
Patients.— RAMQ enrollees 70 years and older. The 3677 cases were patients with
a cataract extraction between 1992 and 1994. The 21868 controls were randomly
selected from patients who did not have a diagnosis of cataract and matched
to cases on the index date of the case.
Main Outcome Measures.— Odds ratio of cataract extraction in patients with prolonged cumulative
exposure to inhaled corticosteroids compared with nonusers.
Results.— Excluding patients with systemic steroid treatment and after adjusting
for age, sex, diabetes, systemic hypertension, glaucoma, ophthalmic steroids,
and the number of physician claims for services, use of inhaled corticosteroids
for more than 3 years was associated with undergoing cataract extraction (odds
ratio [OR], 3.06; 95% confidence interval [CI], 1.53-6.13). For high average
daily doses of beclomethasone or budesonide (>1 mg), the OR was elevated after
more than 2 years of treatment (OR, 3.40; 95% CI, 1.49-7.76), whereas for
low to medium doses (≤1 mg) of these drugs, the OR was 1.63 (95% CI, 0.85-3.13)
after 2 years.
Conclusion.— Prolonged administration of high doses of inhaled corticosteroids increases
the likelihood of undergoing cataract extraction in elderly patients. Further
studies are needed to investigate the risk of developing cataracts for low
to medium doses over longer periods.
INHALED corticosteroids have provided a significant advance in the management
of asthma. Their use is now recommended at a much earlier stage of the disease,
coupled with a trend to prescribe higher doses.1- 3
The increasing use of inhaled corticosteroids over extended periods raises
important questions relating to their safety and the possibility of adverse
For systemic corticosteroids, an increased risk for the development
of posterior subcapsular cataract has been well established.6- 8
A number of case reports also suggested an increased risk of cataract for
inhaled corticosteroids.9- 12
Several clinical studies initiated to investigate this association failed
to find an elevated risk.13- 17
These studies were, however, of limited sample size and mostly conducted in
children, in whom cataract is very rare.
Recently, a cross-sectional study in the elderly reported an increased
risk of cataracts associated with inhaled corticosteroid use.18
Among limitations of their study, the authors discuss the "large amount of
missing data regarding both exposure-related variables and confounders."18 The sample size of the study, although large in comparison
with previous studies, was insufficient to investigate the influence of the
daily dose and duration of inhaled steroid use.19
Consideration of the daily dose is crucial,20
since several studies suggest an increased risk of systemic effects only for
high doses of inhaled corticosteroids.4,5
In this study, we investigate whether the risk of cataract extraction
is associated with the dose and duration of inhaled corticosteroid use and
how the risk compares with that for systemic corticosteroids.
We conducted a case-control study among the elderly population of Quebec
for the years 1987 to 1994, using data from the provincial health insurance
plan database (RAMQ database). This database includes all prescription drugs
and medical services for all individuals 65 years or older. Of the estimated
770925 elderly in Quebec, more than 750000 were registered with the RAMQ in
1990.21 A high level of reliability and validity
of the prescription data has been demonstrated.21
Data in the records include information on the patient's age and sex, diagnoses,
and all filled prescriptions and medical procedures. The prescription data
include the drug name, dispensation date, dose, dosage form, treatment duration,
and quantity of drug dispensed. Drugs dispensed to patients during stays in
hospitals or nursing homes are not included in the database. Diagnoses are
coded according to the International Classification of Diseases,
Ninth Revision (ICD-9).22
The study was conducted using a 10% random sample of the database covering
the years 1987 to 1994.
Cases were subjects with a cataract extraction between 1992 and 1994.
All had at least 5 years of prior follow-up logged in the database. The date
of the cataract extraction was set as the index date for the case. If a patient
had surgery for cataract on more than one occasion, the date of the first
cataract extraction was set as the index date. Controls were randomly selected
among all patients in the database who did not have a diagnosis of cataract
or had a cataract extraction. Like the cases, to be eligible controls were
required to have been enrolled in the database for at least 5 years before
a randomly assigned index date. Potential controls were matched to each case
on the index date of the case to account for seasonal and secular trends in
medication use. Up to 6 controls per case were then randomly selected from
these matched sets.
We identified all dispensed prescriptions for inhaled and oral corticosteroids
for cases and controls before the index date. The following inhaled corticosteroids
are listed on the Quebec provincial drug formulary and were included in the
exposure definition (ie, beclomethasone, budesonide, flunisolide, and triamcinolone).
Oral corticosteroids on the drug formulary included cortisone, hydrocortisone,
prednisone, prednisolone, triamcinolone, methylprednisolone, dexamethasone,
We defined recent exposure to inhaled corticosteroids as a drug dispensation
that lasted into the 14-day period before the index date. We determined the
influence of prolonged treatment with inhaled or oral corticosteroids by calculating
the cumulative treatment duration with these drugs for each patient since
entry into the database. For this calculation, we added the treatment durations
of all dispensations of either inhaled or oral corticosteroids that were filled
before the index date. For both groups of drugs, we defined the following
categories of exposure duration: "no exposure," "exposure of 1 year or less,"
"between 1 to 3 years," and "more than 3 years." To eliminate confounding
by concurrent oral corticosteroid use, we identified all patients who had
not had systemic steroids dispensed before the index date. To reduce contamination
by oral or inhaled corticosteroid use that could have taken place before the
patient entered the database, we identified all patients who had not used
inhaled or systemic steroids during their first year in the database.
To investigate the influence of the daily dose of beclomethasone and/or
budesonide, we determined for all patients the cumulative dose and cumulative
duration of treatment with these drugs from the patients' entry into the database
until the index date. We calculated the average daily dose for each patient
by dividing the cumulative dose of beclomethasone and/or budesonide by the
cumulative treatment duration with these substances. We categorized this measure
of the average daily dose into "1 mg or less" or "more than 1 mg" of beclomethasone
and/or budesonide per day, since doses of beclomethasone and budesonide below
1 mg/d have been suggested to be free of systemic adverse effects in adults.5 To investigate the influence of the dose according
to the duration of use, we categorized the cumulative duration of use of beclomethasone
and/or budesonide into "1 year or less of treatment, "1 to 2 years of treatment,"
and "more than 2 years of treatment."
Covariates included age, sex, systemic hypertension, diabetes mellitus,
glaucoma, the number of physician claims for services in the year before the
index date, and previous use of oral or ophthalmic corticosteroids. Systemic
hypertension was defined by a diagnosis of hypertension before the index date.
Diabetes mellitus was defined by an oral hypoglycemic drug or insulin being
dispensed before the index date. We defined glaucoma by either a diagnosis
of or medical treatment for glaucoma, including therapy with ocular β-blockers,
ocular parasympathomimetics, ocular α-agonists, and carboanhydrase inhibitors.
Because of coding in the database, this definition also includes patients
with ocular hypertension without manifest glaucoma. We classified patients
according to ophthalmic steroid use (any previous use vs no use) and oral
steroid use (use of oral steroids for more than 365 days vs 365 days or less).
We categorized the number of physician claims for services into 10 claims
or less vs more than 10 claims in the year before the index date.
The rate ratio of cataract extraction for each exposure group was estimated
from odds ratios (ORs) calculated by conditional logistic regression using
the SAS PHREG program (SAS Institute Inc, Cary, NC).23
We constructed individual models characterizing patients according to recent
exposure to inhaled corticosteroids, the cumulative duration of inhaled corticosteroid
treatment, and the dose and duration of beclomethasone and/or budesonide treatment.
These analyses were either adjusted for oral corticosteroid use or restricted
to patients who had not used oral corticosteroids before the index date. In
some analyses, we also excluded patients who had used inhaled corticosteroids
during their first year after entry into the database. We also analyzed the
risk for systemic corticosteroids, using the same cumulative exposure periods
as for inhaled corticosteroids, while in one analysis, we excluded patients
treated with systemic corticosteroids during their first year in the database.
For all these analyses, the reference category was the absence of exposure
to the respective form of corticosteroid after entry into the database. All
models simultaneously controlled for all covariates listed above. Two-tailed P values less than .05 were considered significant and
95% confidence intervals (CIs) were calculated for all relative risks. We
further calculated the excess risk (ER) of cataract extraction by determining
the baseline incidence rate (I0) in our study sample and applying
the OR as an estimator of the risk ratio (RR) to this rate by using the following
We identified 10214 patients with cataract extractions, of whom 6537
patients did not meet the eligibility criterion of being in the database for
at least 5 years before the index event, leaving a group of 3677 cases. We
selected 21868 controls who fulfilled the matching and eligibility criteria,
that is an average of 5.9 controls per case. The mean (SD) follow-up period
in the database was similar for cases and controls, with 2354 days (SD, 309)
for cases and 2308 days (SD, 305) for controls.
Characteristics of cases and controls are summarized in Table 1. The risk of cataract extraction increased with age and
was elevated in women, in patients with diabetes, in patients with glaucoma,
and in patients who had been dispensed ophthalmic or oral steroids before
the index date. For patients who had more than 10 physician claims for services
in the year before the index date, the risk was increased 4-fold compared
with patients who had 10 or fewer physician claims.
Table 2 shows that, anytime
during the observation period, 428 cases and 1946 controls were exposed to
inhaled corticosteroids. Beclomethasone and budesonide accounted for the majority
of prescriptions. Seventy-six cases and 278 controls had used more than 1
type of inhaled corticosteroid. A total of 163 cases and 581 controls had
been exposed recently to inhaled glucocorticoids, yielding an adjusted OR
of 1.28 (95% CI, 1.05-1.55).
Table 3 displays the risk
estimates for inhaled corticosteroids associated with different cumulative
periods of use. Excluding patients with systemic corticosteroid use, there
was no increase in risk observed in patients who had used inhaled glucocorticoids
for 1 year or less and in patients whose cumulative treatment duration ranged
from 1 to 3 years. Patients who had used inhaled steroids for more than 3
years had an elevated risk with an adjusted OR of 3.06 (95% CI, 1.53-6.13).
A further slight increase in risk was observed when the analysis excluded
patients who had used inhaled steroids during their first year after entry
into the database.
After excluding patients with systemic steroid use, we identified 227
cases and 1202 controls who had beclomethasone and/or budesonide dispensed
(Table 4). After patients were
stratified by the cumulative treatment duration, cataract extraction was more
frequent among patients who had received high doses of beclomethasone or budesonide
for more than 2 years (adjusted OR, 3.40; 95% CI, 1.49-7.76). The OR for low
to medium doses of beclomethasone or budesonide given for more than 2 years
was 1.63 (95% CI, 0.85-3.13). For less than 2 years of treatment, there was
no increase in risk seen for either low or high doses.
A total of 416 cases and 1724 controls were exposed to oral glucocorticoids
before the index date. Table 5
shows the risk estimates for oral corticosteroids associated with different
cumulative treatment durations. In comparison with inhaled corticosteroids,
the risk of cataract extraction was already increased after more than 1 year
of oral steroid use. A further increase in risk was observed for a cumulative
exposure of more than 3 years. Excluding patients from the analysis who had
used oral corticosteroids during their first year after entry into the database,
as we did in the analysis for inhaled corticosteroids, did not result in important
changes of these risk estimates.
In our study sample, the baseline incidence of cataract extraction was
1.75% per year. Applying our estimate of the rate ratio 3.06 for more than
3 years of inhaled steroid use to this baseline incidence, we calculated an
ER of 3.61% per year, or 361 additional cases per 10000 elderly persons per
The results of our study suggest that prolonged use of inhaled corticosteroids
increases the risk of cataract extraction in the elderly. Patients who had
no use of oral corticosteroids recorded for at least 5 years, but had been
treated with inhaled corticosteroids for more than 3 years, had a 3-fold increased
risk of cataract extraction compared with patients not treated with inhaled
steroids. Based on this risk estimate, we calculated that such extended use
of inhaled corticosteroids would give rise to 361 additional cases of cataract
extraction per 10000 elderly patients per year. Our findings are consistent
with the known relation between use of systemic steroids and cataract formation6,25 and confirm the results of a recent
study from Australia.18 Previous studies investigating
this association did not find an increased risk,13- 17
which may have resulted from insufficient power of these studies.
In the Australian study, Cumming et al18
suggested an increased risk of posterior subcapsular cataract with increasing
lifetime doses of inhaled corticosteroids. Due to the limited sample size
in their study, they could not investigate the risk according to the daily
dose of inhaled steroid and its duration of use.19
In a review of clinical pharmacological studies, it has been suggested that
both beclomethasone and budesonide in doses less than 1 mg/d were free of
systemic adverse effects in adults, while in doses more than 1 mg/d, there
was a potential for systemic effects.5 Classifying
the average daily dose of beclomethasone and budesonide in study patients
according to this cut point into "low to medium" or "high" doses, we did not
observe an increased risk of cataract for either dose category for a treatment
duration of up to 2 years. After more than 2 years of treatment, the risk
was elevated more than 3-fold for high doses, whereas a much smaller and statistically
not significant increase in risk was seen for low to medium doses.
Our study also confirmed an increased risk for various previously described
risk factors for cataract, among them old age, female sex, diabetes mellitus,
glaucoma, and systemic and ophthalmic steroid use.25- 32
For systemic corticosteroids, we observed an increasing risk with increasing
cumulative duration of treatment that already became apparent after treatment
exceeded 1 year. The risk was also considerably elevated for patients with
a higher number of physician encounters, which may reflect an increased risk
of cataract extraction in patients with higher comorbidity. Also, cataract
extraction is an elective procedure, and its timing will also depend on the
Some limitations of the study design need to be considered. We defined
cases by cataract extraction, an end point that has often been used in the
investigation of risk factors for cataract.15,32- 39
Cataracts requiring extraction are those sufficiently severe to affect vision
and, therefore, are of greatest clinical and public health impact. It would
have been desirable to investigate the risk also according to the subtype
of cataract leading to the surgical procedure, but this information was not
available to us. Lack of classification by subtype, however, cannot explain
any of the positive associations we observed. If an association is not common
to all cataract subtypes, then an examination of the subtype-specific risk
will yield even a higher point estimate of the risk than the one presented
in our study.
Although we had for each patient at least 5 years of detailed drug exposure
information preceding the index date, we did not have information on the patients'
drug use before the age of 65 years. The different exposure durations we calculated
may therefore actually be longer when the patients used the medications also
before they entered the database. We also cannot rule out that among the patients
with no systemic corticosteroid use some had used oral glucocorticoids before
the age of 65 years. To address this concern of confounding by previous oral
corticosteroid use, we investigated the risk of cataract extraction in patients
who had not used oral corticosteroids and, in addition, had not used inhaled
corticosteroids during their first year in the database. Since the risk estimate
did not substantially change in this group of patients, which more likely
represents new users of inhaled corticosteroids, confounding by previous systemic
corticosteroid use appears to be of less concern.
Since the database does not contain information about drugs dispensed
in hospitals, we cannot exclude that some patients may have received systemic
corticosteroids while hospitalized. Short treatment courses with systemic
steroids in the hospital will be of minor importance for the development of
cataract, since longer treatment periods with systemic corticosteroids are
usually needed to increase the risk.7 In patients
treated with systemic corticosteroids for an extended period, the corticosteroid
is usually tapered and is likely to be continued after discharge from the
hospital and will, therefore, also appear in the ambulatory records of drug
In our analysis, we controlled for several factors that have been discussed
as risk factors for cataract; however, we did not have information on others,
such as trauma, UV radiation, alcohol, and smoking. We have no reason to believe
that trauma, UV radiation, or alcohol intake are related to exposure with
inhaled corticosteroids and, therefore, do not expect confounding by these
variables. Smoking, described as a risk factor for cataract in some studies,25 but not in others,40,41
is a known risk factor for respiratory diseases. Although it has been suggested
that smokers who develop respiratory symptoms for any reason tend to quit
smoking,42 this may not necessarily be the
case.43 However, since the risk of cataract
in smokers is not high,25 failing to adjust
for smoking should not affect the risk estimates in an important way.
In summary, the results of our study demonstrate an increased risk of
cataract for prolonged use of high doses of inhaled corticosteroids. Further
studies are needed to investigate the risk according to the dose over more
extended treatment periods, since an increased risk also for lower doses of
inhaled steroids cannot be ruled out by our study findings. In addition, in
vitro studies have demonstrated different ratios of topical to systemic activity
for different inhaled steroids4,44;
therefore, studies are also needed that investigate the risk according to
the steroid substance.