CLINICAL SCIENCES

Reliability and Validity of the Ocular Surface

Disease Index
Rhett M. Schiffman, MD, MS; Murray Dale Christianson, MD, FRCSC; Gordon Jacobsen, MS;
Jan D. Hirsch, PhD; Brenda L. Reis, PhD

Objective: To evaluate the validity and reliability of the good to excellent. The OSDI was valid, effectively dis-
Ocular Surface Disease Index (OSDI) questionnaire. criminating between normal, mild to moderate, and se-
vere dry eye disease as defined by both physician’s as-
Methods: Participants (109 patients with dry eye and 30 sessment and a composite disease severity score. The OSDI
normal controls) completed the OSDI, the National Eye also correlated significantly with the McMonnies ques-
Institute Visual Functioning Questionnaire (NEI VFQ- tionnaire, the National Eye Institute Visual Functioning
25), the McMonnies Dry Eye Questionnaire, the Short Questionnaire, the physical component summary score
Form-12 (SF-12) Health Status Questionnaire, and an oph- of the Short Form-12, patient perception of symptoms,
thalmic examination including Schirmer tests, tear breakup and artificial tear usage.
time, and fluorescein and lissamine green staining.
Conclusions: The OSDI is a valid and reliable instru-
Results: Factor analysis identified 3 subscales of the ment for measuring the severity of dry eye disease, and
OSDI: vision-related function, ocular symptoms, and en- it possesses the necessary psychometric properties to be
vironmental triggers. Reliability (measured by Cron- used as an end point in clinical trials.
bach a) ranged from good to excellent for the overall in-
strument and each subscale, and test-retest reliability was Arch Ophthalmol. 2000;118:615-621

D
RY EYE DISEASE is one of To date, the McMonnies Dry Eye
the most frequently Questionnaire is the only patient-
encountered categories perspective instrument specific for dry eye
of ocular morbidity in disease that has a formalized grading
the United States, with scheme and some published psychomet-
as many as 4.3 million persons older than ric properties.4,5 However, this instru-
65 years suffering from symptoms either ment was evaluated as a screening test to
often or all of the time.1,2 The National discriminate subjects with dry eye from a
Eye Institute workshop on clinical trials normal population and not as an instru-
in dry eye defined dry eye as “a disorder ment to grade either dry eye symptom se-
of the tear film due to tear deficiency or verity or its effect on vision-related func-
excessive tear evaporation which causes tion. In addition, formal reliability testing
damage to the interpalpebral ocular sur- on this instrument has not been pub-
face and is associated with symptoms of lished. Bandeen-Roche et al6 developed a
ocular discomfort.” 3 This workshop 6-item symptom inventory for population-
noted that a dry eye condition can exist based epidemiological research. How-
without evidence of ocular surface dam- ever, they did not include a numeric grad-
age and that a primary goal of treatment ing scheme that easily summarizes a
should be to improve symptoms. More- patient’s reported severity. Moreover, the
over, the workshop participants con- reported Cronbach a for this symptom in-
cluded that all clinical trials concerning ventory (0.61) suggests that its internal
dry eye should include an assessment of consistency may not be high enough for
From the Department of Eye
subjective symptoms and functional life- groupwise comparisons.7
Care Services, Henry Ford
Health System, Detroit, Mich style through the use of a well-designed The Ocular Surface Disease Index
(Drs Schiffman and and validated questionnaire, and that (OSDI), developed by the Outcomes Re-
Christianson and Mr Jacobsen), such an instrument may be the best mea- search Group at Allergan Inc (Irvine,
and Allergan Inc, Irvine, Calif sure for determining the clinical efficacy Calif),8 is a 12-item questionnaire de-
(Drs Hirsch and Reis). of therapeutic interventions. signed to provide a rapid assessment of the

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

615

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 SUBJECTS AND METHODS MEASUREMENTS

Health Status Measures

This study was conducted in compliance with the Code of
Federal Regulations for institutional review boards, spon- Participants completed by self-administration the Short
sors, and investigator obligations. Written informed con- Form-12 Health Survey (SF-12), a measure of general health
sent was obtained from all patients before enrollment in status11; the National Eye Institute Visual Functioning Ques-
the study. tionnaire (NEI VFQ-25), a questionnaire found to be reli-
able and valid across several common eye diseases12; the
SUBJECTS McMonnies Dry Eye Questionnaire5; and the OSDI. They
were also asked to record their perception of the severity
Patients 18 years or older who had been diagnosed as hav- of their dry eye symptoms (as measured by a 9-level sub-
ing dry eye symptoms (International Classification of Dis- jective facial expression scale).13 They also completed a medi-
eases, Ninth Revision,9 code 375.15) at the Henry Ford Health cal comorbidity survey.
System in the preceding 12 months and who had had symp-
toms for at least 3 months were recruited for the study. Any Clinical Measures
patients who had undergone temporary or permanent punc-
tal occlusion were eligible 3 months after the procedure. Con- Clinical measures included “walking-around” binocular vi-
trol subjects had to have no significant ocular disease other sual acuity according to the Early Treatment of Diabetic
than refractive error and no systemic disease likely to be as- Retinopathy Study chart; corneal and conjunctival surface
sociated with dry eye. Control subjects were sex-matched and staining using fluorescein and lissamine green, graded ac-
age-matched (± 5 years) to the patients with dry eye disease. cording to the Van Bijsterveld14 and Oxford scoring schemes,
Any patient who had any uncontrolled systemic dis- respectively; tear production according to Schirmer test type
ease or disability that affected his or her activities of daily 1 (with no anesthesia), Schirmer test type 2 (with nasal
living (including ocular allergy, infection, or irritation that stimulation) for those with no tear production on Schirmer
was not related to dry eye disease) was excluded from par- test type 1, and basic secretor’s test (with anesthesia); and
ticipation in the study. Also excluded were patients who had tear quality (using tear breakup time). The presence of
undergone ocular surgery (including cataract surgery) within blepharitis, meibomian gland dysfunction, ocular mucin,
the previous 6 months or who were known to have an al- lid abnormalities, and ocular surface disorders was re-
lergy to any component of any of the agents used in the study, corded as well (based on the physician’s judgment). The
eg, lissamine green, fluorescein, or anesthetic. severity of dry eye disease was assessed in the following 3
All study participants had to have a life expectancy of ways: by physician assessment, by a composite disease se-
1 year or more and “walking-around” visual acuity (with verity score (see below), and by the patients’ frequency of
their usual correction) of 20/40 or better in each eye. They artificial tear use.
had to be English-speaking and cognitively intact (score The composite disease severity score was created to
of $20 on the Folstein Mini-Mental State Examination).10 establish a second measure of severity that was substan-
tially less dependent on the physician’s subjective assess-
STUDY PROTOCOL ment and easily reproducible. It also adheres to recom-
mendations of the National Eye Institute Workshop by
Eligible participants completed a series of questionnaires combining traditional clinical measures of dry eye (Schirmer
in which their sociodemographic status, health status, vi- test type 1 and lissamine green staining) with a symptom-
sual functioning, and ocular symptoms were assessed. Af- based measure (patient perception of ocular symptoms) in
ter completion of the questionnaires, patients underwent the evaluation of dry eye. The criteria for measuring dis-
a detailed ophthalmic examination. Questionnaires were ease severity based on this composite score were set up a
completed before the examination to ensure that the clini- priori and were not modified at any point during the study.
cal encounter would not influence the patients’ responses. First, the patient’s findings were designated as normal, mild
To determine test-retest reliability of the OSDI, all pa- to moderate, or severe for each individual measure. The cut-
tients were asked to return in 2 weeks to complete the ques- off points for the normal and severe groups were based on
tionnaire a second time. values suggested by a review of the medical and scientific

symptoms of ocular irritation consistent with dry eye dis- dry eye condition that affected their daily activities. Item
ease and their impact on vision-related functioning. The responses were then categorized, and categories men-
initial OSDI items were generated from patient com- tioned more than once were formatted into an initial ques-
ments from several years of clinical studies conducted tionnaire. This initial questionnaire included 40 items,
by Allergan Inc, several quality-of-life instruments, and which were later reduced to the final 12 questions on the
suggestions from clinical investigators. This item list was basis of validity and reliability data from 3 groups (2 small
then distributed to more than 400 patients with dry eye groups of patients with dry eye and one phase II clinical
disease, who were asked to indicate whether they expe- trial group).
rienced any of the symptoms or problems on the list and, The objective of the present study was to evaluate
if so, how often. This information was combined with the validity and reliability of the OSDI and to determine
responses from 44 patients with dry eye disease and 2 its usefulness as an end point in clinical trials testing the
health professionals who were asked to list aspects of their efficacy of new treatments for dry eye disease.

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

616

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 literature.13-18 However, there was no consensus or consis- ocular symptoms). Scores were computed for identified sub-
tent trend in the literature indicating appropriate thresh- scales, and these subscales underwent formal reliability and
olds for distinguishing between patients with mild and mod- validity assessment along with the total instrument.
erate values. Therefore, all patients with disease severity
graded as greater than normal but less than severe were com- Reliability
bined into a single mild to moderate group. The severity
designations used for the Schirmer test type 1 were as fol- The reliability of the OSDI questionnaire overall, and for
lows: greater than 10 mm, normal; 6 to 10 mm, mild to mod- the identified subscales within this study, was computed
erate; and 0 to 5 mm, severe. The severity designations used with Cronbach a, a measure of internal consistency. The
for lissamine green staining were as follows: 0 to 1, nor- test-retest reliability of the instrument was evaluated by com-
mal; 2 to 6, mild to moderate; and 7 to 9, severe. The se- puting a random-effects intraclass correlation among pa-
verity designations used for the subjective facial expres- tients who completed the OSDI a second time.
sion scale were as follows: 1 to 3, normal; 4 to 7, mild to
moderate; and 8 to 9, severe. A value of 1 was assigned to Validity
a normal grade for each measure, 2 for a mild to moderate
grade, and 3 for severe. The values for the 3 measures were Discriminant validity of the OSDI was assessed by testing
then summed to generate a final severity score: less than for significant differences in OSDI scores by disease sever-
3, normal; 4 to 7, mild to moderate; and 8 to 9, severe. ity (based on both the physician’s assessment of severity
and the composite disease severity score) by an analysis of
OSDI Scoring Algorithm variance and the Tukey test for multiple comparisons, which
maintains an overall P value of .05. Multiple linear regres-
The 12 items of the OSDI questionnaire were graded on a sion was also used to test for an association between dis-
scale of 0 to 4, where 0 indicates none of the time; 1, some ease severity and OSDI scores while adjusting for factors
of the time; 2, half of the time; 3, most of the time; and 4, considered likely to influence functional status measures,
all of the time. The total OSDI score was then calculated including sociodemographic factors (age, race, sex, edu-
on the basis of the following formula: OSDI = [(sum of scores cation, income, and employment status) and number of
for all questions answered) 3 100]/[(total number of ques- medical comorbidities.
tions answered) 3 4]. Concurrent validity was computed by correlating OSDI
Thus, the OSDI is scored on a scale of 0 to 100, with scores with the other measures of health status (McMon-
higher scores representing greater disability. Subscale scores nies questionnaire, NEI VFQ-25, and SF-12). In addition,
are computed similarly with only the questions from each correlations with the traditional objective measures of dry
subscale used to generate its own score. eye disease (including lissamine green and fluorescein stain-
ing, tear breakup time, Schirmer scores, and artificial tear
STATISTICAL ANALYSIS usage) in the more severely affected (worse) eye were com-
puted with the Spearman coefficient, a nonparametric sta-
The distribution of age, race, sex, income, and educa- tistic, as some of the measures were ordinal.
tional level between patients with dry eye disease and con- Finally, receiver operating characteristic (ROC) curves
trol patients was compared by t test, x2 test, or Fisher ex- were generated to describe the sensitivity and specificity
act test, where appropriate. A P value of .05 was considered of the OSDI for the diagnosis of dry eye at each OSDI score
significant. Mean functional status scores were computed with the use of both the physician’s assessment of severity
for all disease severity categories and the control group. The and the composite disease severity score.20
proportion of patients whose data were at the ceiling and A target enrollment of 150 patients was selected to de-
floor in each category was also computed. tect a correlation coefficient of 0.23, detectable with a power
of 0.80 at an a level of .05. For estimating test-retest reli-
Factor Analysis ability, a sample size of 50 patients was deemed sufficient
to provide a 1-sided 95% confidence interval of 0.84 (lower
Factor analysis with varimax rotation19 was used to deter- bound) for an intraclass correlation of 0.90.21 However, to
mine whether items within the OSDI tended to cluster to- increase the power of the estimate of test-retest reliability,
gether to create subscales (eg, visual functioning and all patients were asked to return for a second test.

RESULTS eye had more associated medical and concomitant ocu-

lar disorders than the control patients (Table 1).
PATIENT CHARACTERISTICS All 139 subjects who were enrolled in the study com-
pleted the first visit; 76 subjects returned for a retest visit.
There were 109 patients with dry eye disease and 30 con- Some patients recruited as control subjects were found
trol subjects. There were no statistically significant dif- to have signs and symptoms of dry eye disease on the ba-
ferences between the patient and control groups in the sis of either the physician’s assessment of severity or the
following demographic variables: age, sex, race, educa- composite disease severity score and were analyzed ac-
tion level, employment, and income. The mean ages for cordingly. Two patients recruited as control subjects re-
subjects in the control and dry eye groups were approxi- ported a history of rheumatoid arthritis at the time of the
mately 55 and 58 years, respectively, with the majority examination but were analyzed as controls on the basis
being female and white. As expected, patients with dry of the physician’s assessment and the computed com-

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

617

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 tal OSDI score and the subscales (Table 2) and once again
Table 1. Demographics and Concurrent Medical met or exceeded 0.7.
and Ocular Conditions
VALIDITY OF OSDI
Dry-Eye Control
Group Group
(n = 109) (n = 30) P When patients were grouped according to the physi-
Age, mean ± SD, y 57.5 ± 13.9 54.5 ± 13.4 .30
cian’s assessment of severity, the mean OSDI total score
Sex, No. (%) F 86 (78.9) 28 (83.3) .59 was 36 in the severe group, 21 in the mild to moderate
Race, No. (%) white 76 (69.7) 20 (66.7) .75 group, and 10 in the control group ( Figure , left;
Associated medical Table 3). The proportion of patients with scores at the
conditions, No. (%) ceiling and floor in each category can also be seen in Table
None 79 (72.5) 25 (93.3) .02
3. All between-group differences were statistically sig-
Sjögren disease 8 (7.3) 0 .20
Rheumatoid arthritis 17 (15.6) 2 (6.7) .37
nificant (P#.05; Tukey test for multiple comparisons)
Systemic lupus erythematosus 3 (2.8) 0 ..99 with the exception of the comparison between the nor-
Progressive systemic sclerosis 1 (0.9) 0 ..99 mal and mild to moderate groups for the vision-related
Polymyositis 1 (0.9) 0 ..99 function subscale, and the mild to moderate and severe
Dermatomyositis 0 0 NA* groups for the environmental triggers subscale.
Concomitant ocular When patients were grouped according to compos-
findings, No. (%)
None 61 (60.0) 23 (76.7) .04
ite disease severity scores, the mean OSDI scores were
Blepharitis 11 (10.1) 1 (3.3) .46 36 in the severe group, 18 in the mild to moderate group,
Meibomian gland dysfunction 24 (22.0) 5 (16.7) .52 and 5 in the control group (Figure, right; Table 3). The
Ocular mucin abnormality 0 0 NA proportion of patients with scores at the ceiling and floor
Disorders of lid aperture 1 (0.9) 1 (3.3) .39 in each category can also be seen in Table 3. All group-
Ocular surface disorder 2 (1.8) 0 ..99 wise comparisons (between the scores for the normal, mild
Other 10 (9.2) 0 .12
to moderate, and severe groups) for the total OSDI score
*Not applicable.
and each subscale of the OSDI were statistically signifi-
cant (P#.05; Tukey test for multiple comparisons).
The OSDI total score and each of the subscale scores
Table 2. Reliability of the Ocular Surface Disease Index
were also significantly associated with disease severity
(as measured by either the physician’s assessment or the
Test-Retest:
composite score) in multivariate analyses adjusting for
Internal Consistency: Intraclass sociodemographic factors (age, race, sex, education, in-
Cronbach a Correlation come, and employment status) and number of medical
(95% Confidence (95% Confidence comorbidities (P#.005).
Interval) Interval)
In comparing disease severity based on the physi-
Ocular Surface Disease Index 0.92 (0.89-0.94) 0.82 (0.73-0.88) cian’s assessment with that determined by the compos-
total score ite score, it was found that 130 of 139 patients were graded
Vision-related function 0.88 (0.84-0.92) 0.70 (0.56-0.80)
Ocular symptoms 0.92 (0.89-0.94) 0.74 (0.62-0.83)
within 1 level of each other. The weighted k, which is a
Environmental triggers 0.78 (0.71-0.84) 0.81 (0.71-0.87) measure of agreement between 2 rating scales,21 was 0.61,
reflecting good agreement. However, patients’ symp-
toms tended to be designated as slightly less severe ac-
cording to the physician’s assessment than they were on
posite score. If these subjects actually had unrecognized the basis of the composite disease severity score.
dry eye, the resulting bias would have been a conserva-
tive one and reduced the observed differences between COMPARISON OF OSDI SCORES WITH
the diseased and control groups. INDIVIDUAL OBJECTIVE CLINICAL MEASURES

FACTOR ANALYSIS The correlation coefficients between the OSDI and tear
OF THE OSDI breakup time, Schirmer test type 1, fluorescein staining,
and lissamine green staining were very low when all sub-
Factor analysis disclosed that there were 3 subscales, in- jects were analyzed together (Table 4). However, when
terpreted as vision-related function (6 questions), ocu- the analysis focused on only patients with Schirmer test
lar symptoms (3 questions), and environmental triggers type 1 scores less than 10 mm, low to moderate statisti-
(3 questions). cally significant correlations were detected for all sub-
scales except vision-related function (Table 4).
RELIABILITY OF THE OSDI In contrast, there was a stronger correlation be-
tween OSDI scores and patient-reported variables. All
The Cronbach a for the overall instrument and each of OSDI scores showed moderate correlations with the Pa-
the subscales ranged from good to excellent (Table 2), tient Perception of Ocular Symptoms that were highly
and all exceeded the 0.7 that is recommended for group statistically significant (P,.001) (Table 5). There was
analyses.7 The intraclass correlation between the test and also a moderate and statistically significant correlation
retest scores was also good to excellent for both the to- between patient-reported artificial tear usage and both

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

618

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 60
Normal (n = 30) Normal (n = 24)
Mild-Moderate (n = 96) Mild-Moderate (n = 83)
50 Severe (n = 13) Severe (n = 31)

40
OSDI Score

30

20

10

0
Total Vision-Related Ocular Environmental Total Vision-Related Ocular Environmental
Function Symptoms Triggers Function Symptoms Triggers

Mean Ocular Surface Disease Index (OSDI) scores (total score and scores for the 3 subscales) by disease severity based on the physician’s assessment (left) and
the composite score (right). Lissamine green scoring was unavailable for 1 patient, preventing the computation of a composite score.

the total OSDI score and all of the OSDI subscales

Table 3. Ocular Surface Disease Index Scores
(P,.001) (Table 5). The mean frequency with which pa-
by Disease Severity
tients used artificial tears ranged from 0 to 9 times a day.
There were also moderate correlations between each No. (%) No. (%)
of the OSDI scores and the McMonnies Dry Eye Ques- Mean ± SD at Ceiling at Floor
tionnaire; all correlations were highly statistically sig- Total Score (Score = 0) (Score = 100)
nificant (P,.001) (Table 5). Physical rating
Both the total OSDI score and the scores for the 3 Normal (n = 30) 9.6 ± 14.2 7 (23.3) 0
subscales were significantly correlated with the NEI Mild-moderate (n = 96) 20.8 ± 20.4 10 (10.4) 1 (1.0)
VFQ-25 score (P,.001) (Table 5). The negative corre- Severe (n = 13) 36.3 ± 16.7 0 0
lations obtained reflect the fact that higher scores on the Composite score*
Normal (n = 24) 4.5 ± 6.6 7 (29.2) 0
OSDI represent greater disease impact, while higher scores
Mild-moderate (n = 83) 18.1 ± 17.1 10 (12.0) 0
on the NEI VFQ-25 reflect less disability. Correlations Severe (n = 31) 36.3 ± 23.1 0 1 (3.2)
between each of the OSDI scores and the NEI VFQ-25
questionnaire were stronger than those between the OSDI *Lissamine green scoring was unavailable for 1 patient, preventing the
and either the McMonnies Dry Eye Questionnaire or the computation of a composite score.
patient perception of ocular symptoms.
The OSDI was also compared with the SF-12, a gen- 1.0 indicating a perfect test for dry eye. The areas under
eral measure of health status (Table 5). There were low to the ROC curve for the OSDI demonstrate good to excel-
moderate correlations between OSDI scores and the physi- lent discrimination with the instrument (Table 6).
cal component summary score of the SF-12 (r = −0.39;
P,.001); however, the correlations between OSDI scores COMMENT
and the mental component summary score were poor.
In the present study, the OSDI demonstrated both high
OPERATING CHARACTERISTICS OF THE OSDI internal consistency (the Cronbach a for the overall in-
strument and each of the subscales ranged from good to
The sensitivity and specificity of the OSDI are shown in excellent) and good to excellent test-retest reliability in a
Table 6. The sensitivity value expresses the proportion large sample of patients with dry eye disease and normal
of patients above the given threshold who have dry eye, controls. The OSDI also demonstrated excellent validity,
while the specificity value expresses the proportion of pa- effectively discriminating between normal, mild to mod-
tients below the given threshold who do not have dry eye. erate, and severe dry eye disease as defined by both the
For the computations presented in Table 6, a threshold physician’s assessment of severity and a composite dis-
was selected that maximized the sum of the sensitivity and ease severity score. Moreover, the OSDI demonstrated good
specificity values. Alternative thresholds could be chosen sensitivity and specificity in distinguishing between nor-
that would selectively maximize either the sensitivity or mal subjects and patients with dry eye disease.
the specificity, depending on how the instrument is to be One of the assumptions of test-retest reliability
used. As expected, the performance of the OSDI im- assessments is that the subject’s condition has remained
proved as it was used to discriminate more severe dis- stable between the time of the first test and the time of
ease. The ROC curves can also be used to describe the per- the retest. However, the symptoms of patients with dry
formance of a test by plotting the sensitivity vs eye disease typically fluctuate, thus violating this
(1 − specificity) for each possible test result. The area un- assumption of stability. This may explain the modest
der the ROC curve can be thought of as a summary mea- decrease in the test-retest reliability compared with
sure for these curves, with 0.5 indicating that the test is Cronbach a and may be unavoidable with this popula-
no better than chance at predicting dry eye disease and tion. Despite this, the test-retest correlation still meets

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

619

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 Table 4. Correlation Between Ocular Surface Disease Index (OSDI) Scores and Individual Clinical Measures in the Worse Eye

Tear Break-up Time Schirmer Type 1 Test Lissamine Green Fluorescein

r P r P r P r P
All Patients
OSDI total score −0.14 .09 −0.21 .01 0.13 .14 0.19 .02
Vision-related function −0.10 .26 −0.14 .10 0.09 .30 0.13 .13
Ocular symptoms −0.15 .09 −0.21 .02 0.12 .16 0.20 .02
Environmental triggers −0.19 .02 −0.18 .04 0.11 .22 0.17 .046
Patients With Dry Eye and Schirmer Test Result ,10 mm
OSDI total score −0.29 .005 −0.38 ,.001 0.29 .005 0.31 .002
Vision-related function −0.20 .05 −0.26 .01 0.19 .07 0.22 .03
Ocular symptoms −0.27 .01 −0.35 ,.001 0.27 .009 0.32 .001
Environmental triggers −0.34 .001 −0.38 ,.001 0.29 .005 0.29 .004

Table 5. Correlation Between OSDI Score and Other Variables*

Patient-Reported Variables

Patient Perception
of Symptoms Other Questionnaires
(Subjective Facial Artificial
Expression Scale) Tear Usage McMonnies NEIVFQ-25 SF-12: Physical SF-12: Mental

r P r P r P r P r P r P
OSDI total score 0.669 ,.001 0.45 ,.001 0.67 ,.001 −0.77 ,.001 −0.39 ,.001 −0.24 .004
Vision-related function 0.483 ,.001 0.36 ,.001 0.52 ,.001 −0.80 ,.001 −0.42 ,.001 −0.16 .07
Ocular symptoms 0.590 ,.001 0.41 ,.001 0.59 ,.001 −0.66 ,.001 −0.34 ,.001 −0.20 .02
Environmental triggers 0.632 ,.001 0.39 ,.001 0.61 ,.001 −0.55 ,.001 −0.22 .01 −0.24 .005

*OSDI indicates Ocular Surface Disease Index; NEI VFQ-25, National Eye Institute Visual Functioning Questionnaire; and SF-12, Short Form-12 Health Survey.

Table 6. Sensitivity, Specificity, and AUC of the OSDI*

Based on Physician Rating Based on Composite Score

TH Sensitivity Specificity AUC TH Sensitivity Specificity AUC

Normal vs all dry eye
OSDI total score 15.0 0.60 0.83 0.73 6.0 0.80 0.79 0.83
Vision-related function 12.5 0.47 0.77 0.62 13.0 0.48 0.88 0.72
Ocular symptoms 33.3 0.39 0.93 0.70 17.0 0.64 0.83 0.78
Environmental triggers 12.5 0.71 0.76 0.76 13.0 0.72 0.91 0.84
Normal vs severe dry eye
OSDI total score 17.0 0.92 0.83 0.91 15.0 0.87 0.96 0.96
Vision-related function 13.0 0.92 0.77 0.85 12.5 0.77 0.88 0.87
Ocular symptoms 17.0 0.92 0.67 0.88 16.7 0.87 0.83 0.91
Environmental triggers 17.0 0.92 0.79 0.89 16.7 0.87 0.91 0.94

*AUC indicates area under the receiver operating characteristic curve; OSDI, Ocular Surface Disease Index; and TH, threshold at which combined sensitivity and
specificity values were at maximum.

or exceeds the level recommended for group compari- eye disease, was designed as a screening test to discrimi-
sons of 0.7.7 nate patients with dry eye disease from a normal popu-
The OSDI scores correlated well with other eye- lation5 and primarily uses dichotomous responses (yes
specific health status measures, such as the McMonnies or no) to assess the presence of symptoms. The NEI
questionnaire and the NEI VFQ-25. The correlation was VFQ-25 surveys general ocular health18 and is not in-
not perfect, however, indicating that the OSDI captures tended to capture the broad range of symptoms unique
unique aspects of dry eye disease not addressed by these to a certain ocular disorder.
other instruments. This is to be expected, considering the When the results from all patients were analyzed to-
difference in the contents and structure of these differ- gether, OSDI scores did not correlate well with tradi-
ent questionnaires. The OSDI is specific for dry eye dis- tional objective clinical measures of dry eye (such as
ease and asks patients the frequency of specific symp- Schirmer test type 1). This is consistent with previous
toms and their impact on vision-related functioning. The studies that also failed to find strong correlations be-
McMonnies questionnaire, although also specific for dry tween objective clinical signs of dry eye and patient symp-

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

620

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015
 toms.3 This lack of correlation may be because, among a 3. Lemp MA. Report of the National Eye Institute/Industry Workshop on Clinical
Trials in Dry Eyes. CLAO J. 1995;21:221-232.
heterogeneous group of patients with dry eye, these mea-
4. McMonnies CW, Optom AH. Patient history in screening for dry eye conditions.
sures lack sufficient sensitivity to capture the full range J Am Optom Assoc. 1987;58:296-301.
of ocular surface and tear abnormalities that produce typi- 5. McMonnies CW, Optom AH. Responses to a dry eye questionnaire from a nor-
cal dry eye symptoms. However, in certain subsets of pa- mal population. J Am Optom Assoc. 1987;58:588-589.
tients, these clinical measures may correlate more closely 6. Bandeen-Roche K, Munoz B, Tielsch JM, West SK, Schein OD. Self-reported as-
sessment of dry eye in a population-based setting. Invest Ophthalmol Vis Sci.
with patient perception of disease severity. In support of 1997;38:2469-2475.
this, the present study found that the OSDI correlated 7. Nunnally J. Psychometric Theory. 2nd ed. New York, NY: McGraw-Hill Book Co;
moderately with clinical signs among patients with dry 1978.
eye disease who had tear deficiency. 8. Walt JG, Rowe MM, Stern KL. Evaluating the functional impact of dry eye: the
Ocular Surface Disease Index [abstract]. Drug Inf J. 1997;31:1436.
The OSDI is, therefore, a unique instrument able to
9. World Health Organization. International Classification of Diseases, Ninth Revi-
assess both the frequency of dry eye symptoms and their sion (ICD-9). Geneva, Switzerland: World Health Organization; 1977.
impact on vision-related functioning. The OSDI has good 10. Folstein MF, Folstein SE, McHugh PR. “Mini-Mental State”: a practical method
to excellent reliability, validity, sensitivity, and specificity. for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;
It should prove a valuable complement to other clinical and 12:189-198.
11. Ware JE, Kosinski M, Keller SD. SF-12: How to Score the SF-12 Physical and
subjective measures of dry eye disease by providing a quan- Mental Summary Scales. Boston, Mass: Health Institute, New England Medical
tifiable assessment of dry eye symptom frequency and the Center; 1995.
impact of these symptoms on vision-related functioning. 12. Mangione CM, Lee PP, Pitts J, Gutierrez P, Berry S, Hays RD, for the NEI-VFQ
It also has the necessary psychometric properties to be used Field Test Investigators. Psychometric properties of the National Eye Institute
Visual Function Questionnaire (NEI-VFQ). Arch Ophthalmol. 1998;116:1496-
as an end point in clinical trials of dry eye disease.
1504.
13. Tsubota K. Dry Eye Clinic for Fatigued Eyes [in Japanese]. Tokyo, Japan: Nihon-
Accepted for publication December 10, 1999. jitsugoshuponsha; 1992.
This study was supported by an unrestricted grant from 14. Van Bijsterveld OP. Diagnostic tests in the sicca syndrome. Ophthalmology. 1969;
82:10-14.
Allergan Inc, Irvine, Calif. None of the non-Allergan au- 15. Fox RI, Robinson CA, Curd JG, et al. Sjögren’s syndrome: proposed criteria for
thors have any financial interest in Allergan Inc. the classification. Arthritis Rheum. 1986;29:577-585.
Reprints: Rhett M. Schiffman, MD, MS, Henry Ford 16. Manthorpe R, Anderson V, Jensen OA, et al. Editorial comments to the four sets
Hospital, 2799 W Grand Blvd, Detroit, MI 48202 (e-mail: of criteria for Sjögren’s syndrome. Scand J Rheumatol Suppl. 1986;61:31-35.
[email protected]). 17. Norn MS. Lissamine green: vital staining of cornea and conjunctiva. Acta Oph-
thalmol (Copenh). 1973;51:483-491.
18. Vitali C, Bombarderi S, Haralampos M, et al. Preliminary criteria for the classi-
REFERENCES fication of Sjögren’s syndrome: results of a prospective action supported by the
European Community. Arthritis Rheum. 1993;36:340-347.
19. Hatcher L. A Step-by-Step Approach to Using the SAS System for Factor Analy-
1. Albert DM, Jakobiec FA. Water and electrolyte secretion and fluid modification. sis and Structural Equation Modeling. Cary, NC: SAS Institute Inc; 1994.
In: Principles and Practice of Ophthalmology Basic Sciences [book on CD- 20. Hanley, JA, McNeil BJ. The meaning and use of the area under a receiver oper-
ROM]. Vol 1.1A. Philadelphia, Pa: WB Saunders Co; 1995:sect 1B, chap 29B. ating characteristic (ROC) curve. Radiology. 1982;143:29-36.
2. Schein OD, Munoz B, Tielsch JM, Bandeen-Roche K, West S. Prevalence of dry 21. Fleiss JL. The Design and Analysis of Clinical Experiments. New York, NY: John
eye among the elderly. Am J Ophthalmol. 1997;124:723-728. Wiley & Sons Inc; 1986:8-14.

Notice to Authors: Submission of Manuscripts

Selected manuscripts submitted to the Archives of Ophthalmology will

be submitted for electronic peer review. Please enclose a diskette with
your submission containing the following information:
File name
Make of computer
Model number
Operating system
Word processing program and version number

ARCH OPHTHALMOL / VOL 118, MAY 2000 WWW.ARCHOPHTHALMOL.COM

621

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archopht.jamanetwork.com/ by a Northeastern University Libraries User on 05/26/2015