Validation of a Questionnaire for

Comparing the Tolerability of

Ophthalmic Medications

Beth L. Barber, PhD/ Ellen R. Strahlman, MD, MHS, 2 Robert Laibovitz, MD/
Harry A. Guess, MD, PhD, 2 Scott A. Reines, MD, PhD4

Purpose: The Comparison of Ophthalmic Medications for Tolerability (COMTOL)

questionnaire was developed for use in clinical trials to compare the tolerability of topical
ophthalmic medications used in the treatment of glaucoma.
The questionnaire captures the frequency and bother of common side effects (i.e.,
ocular and other local effects, and effects on visual function) of topical therapy for lowering
intraocular pressure. In addition, the questionnaire measures the extent to which these
side effects and any associated limitations in routine living activities interfere with health-
related quality of life, medication compliance, and patient satisfaction with the medication.
This study was designed to assess the measurement characteristics of the COMTOL
questionnaire.
Methods: The internal consistency, reliability, reproducibility, construct validity, dis-
criminant validity, and responsiveness of the questionnaire were assessed in 70 adult
patients with glaucoma in a clinical trial comparing timolol and pilocarpine.
Results: The questionnaire showed good-to-excellent internal consistency (0.73 to
0.98), reliability (0.76 to 0.94}, and reproducibility (0.75 to 0.93). In general, there was a
strong correlation in the expected direction between the frequency and bother of side
effects and patient-perceived global measures. The questionnaire discriminated between
patients receiving timolol and patients receiving pilocarpine. The questionnaire demon-
strated significant responsiveness to change.
Conclusions: The COMTOL questionnaire showed acceptable measurement char-
acteristics for inclusion as a tolerability measure to supplement spontaneous adverse
event reporting in clinical trials of topical ophthalmic therapy.
Ophthalmology 1997; 104:334-342

Understanding the potential side effects of medications
and their impact on health-related quality of life is im-
portant to patients. This is particularly true for patients
Originally received: June 19, 1995.
Revision accepted: September 20, 1996.
1
Department of Outcomes and Performance Assessment, Merck-Medco
Managed Care, Montvale, New Jersey.
2
Departments· of Epidemiology (ES and HG), and Clinical Neurosci-
ences (SR), Merck Research Laboratories, West Point, Pennsylvania.
3
Eye Research Associates, Austin, Texas.
4
Department of Clinical Neurosciences, Merck Research Laboratories,
West Point, Pennsylvania.
Presented at the ARVO Annual Meeting, Ft. Lauderdale, April 1995.
Supported by Merck Research Laboratories, West Point, P A.
Reprint requests to Beth L. Barber, PhD, Merck-Medco Managed Care,
Inc., 100 Summit Ave, Montvale, NJ 07645-1753.
who have glaucoma, because the disease is chronic and
several medications are required in many patients. Cur-
rently, timolol is the most commonly prescribed ocular
hypotensive agent for the treatment of open-angle glau-
coma. Although timolol is generally well tolerated, sec-
ond-line agents (pilocarpine, epinephrine, and oral car-
bonic anhydrase inhibitors) have less desirable tolerability
profiles.
To compare different treatments for glaucoma by char-
acterizing their common side effects and the health-re-
lated quality of life of patients receiving treatment, we
developed an evaluative questionnaire for Comparing
Ophthalmic Medications for Tolerability (COMTOL).
The questionnaire is designed to capture the frequency
and bother of common side effects of topical drugs used
to control intraocular pressure (lOP); the extent to which
these side effects and any associated limitations in routine

334
 Barber et al · COMTOL Validation Study
living activities interfere with the patients' health-related
quality of life, their compliance with the medication, and
their satisfaction with the medication. The questionnaire
is intended to be administered by an interviewer. We
assessed the measurement characteristics of the question-
naire to ensure that it correctly measures common side
effects and their impact on quality of life, and that it
can be used as an evaluative instrument. This assessment
included measuring internal consistency, test-retest relia-
bility, reproducibility, construct validity, discriminant va-
lidity, and responsiveness. This paper describes the ques-
tionnaire, defines domains and global questions, and pres-
ents their measurement characteristics.
Methods
Patients
The 70 study participants were men and women, age 18 or
older, with open-angle glaucoma or ocular hypertension.
Patients were excluded from the study if they had pre-
viously received treatment with pilocarpine. Patients were
also excluded from the study for ocular conditions includ-
ing: visual acuity less than 20/80 in both eyes, history or
evidence of acute or chronic angle closure glaucoma, pu-
pil unable to dilate sufficiently for an adequate retinal
examination, history or presence of uveitis, retinal detach-
ment, or any other retinal condition for which treatment
with pilocarpine might be inappropriate. Patients were
also excluded from the study who had any contraindica-
tion to the use of timolol or pilocarpine ophthalmic solu-
tion, a history of asthma or chronic obstructive pulmonary
disease (COPD), clinically significant renal disease, or
severe physical disabilities.
Study Design
This 2-week, randomized, parallel study assessed the
measurement characteristics of a questionnaire designed
to compare the side effects of different topical ophthalmic
solutions for lowering lOP, and how these side effects
affected the health-related quality of life of patients who
received these medications. A three-week saline run-in
period, which included a washout of ocular hypotensive
medications, preceded randomization to either timolol
0.5% twice a day or pilocarpine 2% four times a day.
Patients were examined at pretreatment and on study days
8, 13, and 15. The questionnaire was administered by an
interviewer at baseline (day 1) and on study days 13 and
15.
Because the purpose of the study was to validate a
questionnaire designed to compare topical ophthalmic
agents, and because the dosing regimen is a part of that
assessment, treatment regimens were deliberately not
masked to the patient or ophthalmologist examiner. How-
ever, in order to obtain the most objective responses possi-
ble, the questionnaire interviewer was masked to the pa-
tient's regimen, the interviews were conducted in a room
with bright lights (to induce miosis in all patients), and
the patients were instructed not to disclose their dosing
frequency to the interviewer.
Measurements
Clinical Assessments. Clinical assessments, including vi-
sual acuity, slit-lamp examination, lOP measurements (2
hours after the morning medication), and ophthalmoscopy
through dilated pupils, were performed at baseline and
on study days 8, 13, and 15.
COMTOL Questionnaire. The Comparison of Oph-
thalmic Medications for Tolerability (COMTOL) ques-
tionnaire (Appendix A) was interviewer-administered at
baseline and on study days 13 and 15. It consisted of
twelve questions that were based on common side effects
reported by patients in clinical trials of therapy for low-
ering lOP.
Questions 1 and 2 (preference and reason for prefer-
ence), are intended to be asked only at the conclusion
of a cross-over trial, and hence were not asked or as-
sessed in this validation study (which was a parallel
study).
Question 3 establishes which side effects (from a list
of 15 possible side effects) the patient has experienced.
This information is not intended to be analyzed, but
instead to facilitate asking questions 4 and 5, which
make sense only if the patient has experienced the side
effect.
Question 4 refers to each side effect in question 3
that the patient reported having, and asks about the
frequency with which the patient experienced the side
effect.
Question 5 refers to each side effect identified in ques-
tion 3, and asks about its bothersomeness.
Question 6 asks about the extent to which the patient's
quality of life has been interfered with by these side
effects.
Question 7, which is asked only at baseline, identifies
activities the patient performs on a routine basis.
Question 8 refers to each activity identified by the
patient in question 7, and asks the extent to which the
patient had to limit the activity as a result of using the
eye drops.
Question 9 asks the extent to which the patient's qual-
ity of life has been interfered with by these activity
limitations.
Question 10 asks how often the patient missed one or
more doses of test medication.
Question 11 asks those patients who reported missing
at least one dose (in question 10) to indicate why they
did not take the eye drops.
Question 12 asks how satisfied the patient has been
with the test medication.
Defining and Scoring Domains and Global Ques·
tions. Question 4 was scored from 0 ("I did not have
the symptom") to 6 ("Always"). Question 5 was scored
335
 Ophthalmology Volume 104, Number 2, February 1997
Table 1. Baseline Characteristics
Timolol Group
(N = 35)
Age (yrs)
Mean 52.2
(Range) (36-75)
Gender
Female 20
Male 15
Race
White 32
Black 2 1
Asian 1 1
Hispanic 0 2
Intraocular pressure (mmHg)
Mean (SD) 24.54 (3.05)
SO = standard deviation.
from 0 ("Not at all bothered") to 5 ("Extremely both-
ered"). Questions 6 and 9 were scored 0 ("Not at all")
to 5 ("Extremely"). Question 8 was scored 0 ("Not at
all) to 6 ("Extremely"). Question 10 was scored 0 ("I
did not miss any doses") to 6 ("Always"). Question 11
was scored 0 ("I did not take the eye drops because of
side effects that I experienced") or 1 ("I did not take the
eye drops for some other reason."). Question 12 was
scored 0 ("Totally satisfied") to 5 ("Totally dissatis-
fied"). Thus, with all questions except question 11, the
higher score indicated the patients' increased discomfort.
Factor analysis using the varimax rotation of the princi-
pal components procedure' was performed with all items
in question 4 for each treatment group to establish mean-
ingful underlying domains. This analysis was repeated
for all items in questions 5 and 8. Redundant items and
"routine" activities that were not performed by more
than half of the patients were eliminated. Domain scores
were calculated as the average of the items in the domain.
Questions 6, 9, 10, 11, and 12 were prespecified global
questions, which were scored as indicated above.
Measurement Characteristics. A comprehensive eval-
uation of the measurement characteristics of an evaluative
instrument should address internal consistency, test-retest
reliability, reproducibility, construct validity, discrimi-
nant validity, and responsiveness.
We measured internal consistency with Cronbach' s
alpha. 2 Good internal consistency is defined as an alpha
coefficient greater than 0.70 and less than 0.95. 3 Reliabil-
ity refers to the extent to which a single observation can
consistently distinguish between members of a popula-
tion. We used the intraclass correlation coefficient (ICC) 4
to quantify reliability. An ICC of 0.75 and above is con-
sidered acceptable for a research instrument. 5
Reproducibility refers to the extent to which repeated
administration to a stable population yields the same re-
sults. We used the concordance correlation coefficient6 to
quantify reproducibility.
The underlying construct that the symptoms measure
336
Pilocarpine Group All Patients
(N = 35) (N = 70)
54.7 53.4
(33-82) (33-82)
23 43 (61 %)
12 27 (39%)
31 63 (90%)
3 (4%)
2 (3%)
2 (3%)
23.37 (3.65) 23.96 (3.39)
is that of commonly reported side effects of lOP therapy
being frequent and bothersome, causing limitations in liv-
ing activities, and adversely affecting patients' self-per-
ceived quality of life. We assessed construct validity by
correlating side effect and living activity domain scores
with four patient-perceived global measures: (i) interfer-
ence of side effects on overall quality of life, (ii) interfer-
ence of activity limitations on overall quality of life, (iii)
compliance with therapy, and (iv) satisfaction with ther-
apy. Correlations exceeding twice their estimated stan-
dard error were considered to be statistically significant.
We assessed discriminant validity by evaluating the
ability of the questionnaire to distinguish between the
scores of patients receiving timolol and the scores of pa-
tients receiving pilocarpine. For these comparisons, co-
variance was analyzed, comparing least square mean
scores at day 15 while controlling for baseline scores.
Responsiveness is the extent to which scores change
when subjects improve or deteriorate. Changes (from
baseline to day 15) in side effect and living activity do-
main scores in the timolol group were compared (using
Student's t tests) to changes in side effect and living
activity domain scores in the pilocarpine group to deter-
mine whether differences between the two groups could
be detected.
All 95% confidence interval limits for Cronbach's
alphas, intraclass correlation coefficients, and concor-
dance correlation coefficients, were calculated using the
Jackknife method. 7
Results
A total of 70 patients from one center was entered into
this study. The predominantly white (90%) population
had an average age of 53 years (range 33 to 82) and
was 61% female (Table 1). Treatment groups were very
similar with respect to baseline characteristics.
Six patients discontinued therapy prior to completing
 Barber et al · COMTOL Validation Study

Table 2. Domains
Domain Items in Domain Scoring

Frequency of ocular symptoms Burning/stinging, redness, itchy 0 (did not experience) to

Frequency of taste side effects
Frequency of vision
difficulties
Frequency of accommodation
difficulties
Frequency of brow ache
Bothersomeness of ocular
symptoms
Bothersomeness of taste side
effects
Bothersomeness of vision
difficulties
Bothersomeness of
accommodation difficulties
Bothersomeness of brow ache
Limitations on driving
Limitations on reading
Limitations on moderate
activities
eyes, discharge, swelling, dry
eyes, tearing
Bitter taste, unusual taste
Blurred vision, dimming of vision,
trouble seeing at night
Trouble reading, trouble focusing
near to far
Brow ache
Burning/stinging, redness, itchy
eyes, discharge, swelling, dry
eyes, tearing
Bitter taste, unusual taste
Blurred vision, dimming of vision,
trouble seeing at night
Trouble reading, trouble focusing
near to far
Brow ache
Driving at night, driving during
the day
Reading the newspaper, reading
other than the newspaper
Lifting or carrying groceries,
climbing 1 flight of stairs,
walking several blocks
6 (always) ·
0 (did not experience) to
6 (always)
0 (did not experience) to
6 (always)
0 (did not experience) to
6 (always)
0 (did not experience) to
6 (always)
0 (not at all bothered) to
5 (extremely bothered)
0 (not at all bothered) to
5 (extremely bothered)
0 (not at all bothered) to
5 (extremely bothered)
0 (not at all bothered) to
5 (extremely bothered)
0 (not at all bothered) to
5 (extremely bothered)
0 (not at all limited) to
5 (extremely limited)
0 (not at all limited) to
5 (extremely limited)
0 (not at all limited) to
5 (extremely limited)

the study due to clinical adverse effects. However, all 70 velop and validate a questionnaire that would be general-
patients responded to the questionnaire at the three time izable for comparison of other topical ophthalmic agents
points. Five of the six patients discontinued therapy while (not simply timolol and pilocarpine). Therefore, because
receiving pilocarpine due to drug-related adverse experi- some topical ophthalmic agents may cause severe accom-
ences that were possibly, probably, or definitely drug-
related. The sixth patient discontinued therapy while re-
ceiving timolol after experiencing headache, insomnia,
Table 3. Internal Consistency
and skin rash that were possibly drug related.
Cronbach's 95% Confidence
Domain Definitions Domain Alpha Interval*

Factor analysis using the varimax rotation of the principal Frequency of

Ocular symptoms 0.73 (0.59, 0.83)
components procedure was performed with all items for Taste side effects 0.85 (0.01, 0.99)
each treatment group in order to establish meaningful Vision difficulties 0.80 (0.68, 0.88)
underlying domains. As might be expected, the groupings Accommodation difficulties 0.89 (0. 74, 0.95)
were not identical for each treatment group because the Bothersomeness of
two drugs cause somewhat different side effects of differ- Ocular symptoms 0.82 (0.70, 0.90)
ent intensities. From the 15 items comprising question 4 Taste side effects 0.96 (0.17, 0.99)
(frequency of side effects), three groupings arose for the Vision difficulties 0.86 (0.75, 0.93)
pilocarpine treatment group: ocular symptoms, taste, and Accommodation difficulties 0.90 (0.78, 0.96)
vision problems (which included items for difficulties Limitations on
Driving 0.95 (0.87, 0.98)
with vision, accommodation, and brow ache [Table 2]).
Reading 0.98 (0.90, 0.99)
The groupings for the timolol group were slightly differ- 0.98 (0.86, 0.99)
Moderate activities
ent. The tearing item was grouped with the vision diffi-
culty items. Although the factor analysis suggested three * 95% Confidence interval estimates are based on the jackknife.
domains, an objective of this validation study was to de-

337
 Ophthalmology Volume 104, Number 2, February 1997

Table 4. Test-Retest Reliability and Reproducibility

lntraclass Correlation Concordance Correlation
Domain N (95% Confidence Interval)* (95% Confidence Interval)*
Frequency of
Ocular symptoms 70 0.84 (0.59, 0.93) 0.83 (0.59, 0.93)
Taste side effects 70 0.87 (0.40, 0.98) 0.86 (0.40, 0.98)
Vision difficulties 70 0.85 (0.63, 0.94) 0.85 (0.64, 0.94)
Accommodation difficulties 70 0.77 (0.56, 0.88) 0.77 (0.56, 0.88)
Brow ache 70 0.84 (0.50, 0.94) 0.84 (0.50, 0.94)
Bothersomeness of
Ocular symptoms 70 0.85 (0. 79, 0.90) 0.85 (0. 79, 0.90)
Taste side effects 70 0. 75 ( -0.82, 0.99) 0. 74 ( -0.82, 0.99)
Vision difficulties 70 0.91 (0.85, 0.95) 0.91 (0.85, 0.95)
Accommodation difficulties 70 0.76 (0.48, 0.89) 0. 75 (0.48, 0.89)
Brow ache 70 0.86 (0.31, 0.97) 0.86 (0.30, 0.97)
Limitations on
Driving 67 0.88 (0.56, 0.96) 0.88 (0.56, 0.96)
Reading 67 0.91 (0.81, 0.96) 0.91 (0.81, 0.96)
Moderate activities 68 0.92 (0.65, 0.98) 0.92 (0.65, 0.98)
Global
6. QOL interfered with by side effects? 69 0.90 (0.81, 0.95) 0.90 (0.81, 0.95)
9. QOL interfered with by activity
limitations? 70 0.87 (0.72, 0.93) 0.86 (0.72, 0.93)
10. How often did you miss one or more
doses? 70 0.94 (0.87, 0.97) 0.93 (0.87, 0.97)
11. Why did you not take the eyt; drops?
side effects/other 22 0.83 (0.04, 0.96) 0.82 (0.05, 0.96)
12. How satisfied have you been with
the test medication? 70 0.89 (0.49, 0.97) 0.89 (0.49, 0.97)

QOL = quality of life.

* 95% Confidence interval estimates are based on the jackknife.

modation problems and less severe vision problems, and
some other drugs may cause brow ache without vision
problems, problems with vision, accommodation, and
brow ache were defined as three separate domains provid-
ing a total of five frequency-of-side-effects domains (Ta-
ble 2).
Five similar domains were defined for the 15 items
regarding bothersomeness of side effects (question 5):
ocular symptoms, taste, vision, accommodation, and brow
ache. Three limitation-of-living-activities domains were
defined using 7 of the 14 items that comprised question
8 (the other 7 items were eliminated: see Item Reduction,
below). In the future, when the COMTOL is administered
in a clinical trial, patients will be asked at baseline to
identify from the list of 7 items the activities they perform
on a routine basis.
Item Reduction
All side effects were retained in the questionnaire, even
if they occurred infrequently in patients receiving timolol
or pilocarpine.
Items on "routine activities" (e.g., "vigorous activi-
ties'' and ''climbing more than one flight of stairs'') were
eliminated if less than half the patients chose the item as
a routine activity at baseline. Other items were eliminated
if they were redundant. For instance, the items ''walking
more than I mile," "walking several blocks," and
''walking 1 block'' all seem to measure the same patient
capability; therefore, because the items ''walking more
than 1 mile" and "walking I block" were both selected
by fewer patients than the item ''walking several blocks,''
only the item "walking several blocks" was kept. The
item ''bathing or dressing yourself' was eliminated be-
cause the average scores for this item at day I5 were the
smallest averages of all items (timolol group average =
0.00, pilocarpine group average = 0.40), indicating that
topical ophthalmic drugs are unlikely to significantly limit
these activities.
Internal Consistency
The Cronbach' s alphas for all domains with more than
one question had high values (0.73-0.98), supporting the
domain definitions (Table 3). This statistic is not defined
for domains consisting of only one question.
Reliability and Reproducibility
The intraclass correlation coefficients (0.75 to 0.94) and
the concordance correlation coefficients (0.74 to 0.93) for

338
 Barber et al · COMTOL Validation Study

Table 5. Construct Validity Using Correlation Analysis: Global Question 6-

Extent to Which Quality of Life Was Interfered with by Side Effects
Domain N Timolol Group N Pilocarpine Group
Frequency of
Ocular symptoms 34 0.257 35 0.409*
Taste side effects 34 0.888* 35 0.262
Vision difficulties 34 0.748* 35 0.549*
Accommodation difficulties 34 0.318 35 0.638*
Brow ache 34 0.760* 35 0.638*
Bothersomeness of
Ocular symptoms 34 0.370* 35 0.493*
Taste side effects 34 -0.050 35 0.268
Vision difficulties 34 0.849* 35 0.780*
Accommodation difficulties 34 0.417* 35 0.826*
Brow ache 34 0.830* 35 0.777*
Limitations on
Driving 33 0.967* 35 0.887*
Reading 34 0.958* 35 0.886*
Moderate activities 34 0.192 35 0.565*

* Correlations exceeding twice their estimated standard error.

the 13 domains and 4 global questions indicate good-to-
excellent reliability and reproducibility (Table 4).
Construct Validity
Correlations between global questions 6 and 9 and domain
scores were strong and in the expected direction, with the
exception of the bothersomeness-of-taste domain (Tables
5 and 6).
We also correlated domains with global questions 10
and 12, although tables are not presented. Global question
10 (missing one or more doses) did not correlate signifi-
cantly with any domain for timolol. However, for pilocar-
pine, this global did correlate with two of the bother-
sameness domains (vision and accommodation), and with
all three limitation-with-living-activities domains (driv-
ing, reading, and moderate activities). Global question 12
(satisfaction with medication) correlated strongly with all
domains (except taste) for pilocarpine, and correlated
moderately with four domains (frequency and bother-

Table 6. Construct Validity Using Correlation Analysis: Global Question 9-

Extent to Which Quality of Life Was Interfered with by Activity Limitations

Domain N Timolol Group N Pilocarpine Group

Frequency of
Ocular symptoms 35 0.287* 35 0.425*
Taste side effects 35 0.941 * 35 0.362*
Vision difficulties 35 0.794* 35 0.457*
Accommodation difficulties 35 0.365* 35 0.551 *
Brow ache 35 0.690* 35 0.579*
Bothersomeness of
Ocular symptoms 35 0.219 35 0.504*
Taste side effects 35 -0.036 35 0.412*
Vision difficulties 35 0.822* 35 0.714*
Accommodation difficulties 35 0.412* 35 0.752*
Brow ache 35 0.782* 35 0.750*
Limitations on
Driving 34 0.921 * 35 0.844*
Reading 35 0.927* 35 0.807*
Moderate activities 35 0.216 35 0.657*

* Correlations exceeding twice their estimated standard error.

339
 Ophthalmology Volume 104, Number 2, February 1997

Table 7. Discriminant Validity Using Analysis of Covariance

Timolol Group Pilocarpine Group Pr > ITI
LSMEAN LSMEAN H.,: LSMEAN (Timolol) =
Domain Response Response LSMEAN (Pilocarpine)
Frequency of
Ocular symptoms 0.63 1.60 0.0001 *
Taste side effects 0.19 0.36 0.3218*
Vision difficulties 0.55 3.23 0.0001 *
Accommodation difficulties 0.46 3.08 0.0001 *
Brow ache 0.21 2.59 0.0001 *

Bothersomeness of
Ocular symptoms 0.23 0.87 0.0014*
Taste side effects 0.09 0.20 0.1319
Vision difficulties 0.26 2.00 0.0001 *
Accommodation difficulties 0.21 1.85 0.0001 *
Brow ache 0.11 1.69 0.0001 *

Limitations on
Driving 0.15 1.68 0.0001 *
Reading 0.16 1.63 0.0001 *
Moderate activities 0.01 0.49 0.0275*
Global
6. QOL interfered with by side
effects? 0.21 1.65 0.0001 *
9. QOL interfered with by
activity limitations? 0.14 1.37 0.0001 *
10. How often did you miss one
or more doses? 0.45 1.43 0.0106*
11. Why did you not take the
eye drops? side effects/other 1.77 1.50 0.1996
12. How satisfied have you been
with the test medication? 1.17 2.92 0.0001 *

QOL = quality of life.

* Significant at the 0.05 level of significance.

sameness of ocular symptoms and accommodation diffi- Responsiveness

culties) for timolol. In general, there was a greater number
of significant correlations between domains and globals
for pilocarpine than for timolol. This may be due to the
greater frequency and bother of the pilocarpine-induced
side effects.
In almost all cases, the questionnaire was responsive to
changes in scores in the timolol group compared with
changes in scores in the pilocarpine group (Table 8). The
only exceptions were with the frequency and bother-
sameness of taste domains.

Discriminant Validity
The questionnaire clearly discriminated between the pilo-
carpine and timolol groups with respect to: frequency and
bothersomeness of ocular symptoms, vision difficulties,
accommodation difficulties, brow ache; limitations on
driving, reading, and moderate activities; interference-
with-quality-of-life (both questions 6 and 9); frequency
of missed doses (question 10); and satisfaction with medi-
cation (question 12) (Table 7). The questionnaire did not
discriminate between pilocarpine and timolol groups with
respect to frequency and bothersomeness of taste side
effects, or between the two groups with respect to why
eye drops were missed (question 11). For this reason,
question 11 was eliminated from the final validated COM-
TOL questionnaire.
Discussion
We assessed the measurement characteristics of the
COMTOL questionnaire that was designed to distinguish
between different topical ophthalmic medications on the
basis of tolerability. Within the questionnaire, using factor
analysis, clinical judgment, and item importance, we de-
fined:
• Five frequency-of-side-effects domains (ocular
symptoms, taste, accommodation, vision, and brow
ache);
• Five bothersomeness-of-side-effects domains (ocu-
lar symptoms, taste, accommodation, vision, and
brow ache); and

340
 Barber et al · COMTOL Validation Study

Table 8. Responsiveness: Comparing Changes in Scores in the Timolol Group with Changes in Scores in
the Pilocarpine Group
Timolol Group Pilocarpine Group
Change from Baseline Change from Baseline
Baseline Baseline
Mean Mean p
Domain N Mean N Change SD N Mean N Change SD Value
Frequency of
Ocular symptoms 35 0.22 35 0.41 0.72 35 0.25 35 1.36 1.32 0.0004*
Taste side effects 35 0.01 35 0.10 0.53 35 0.10 35 0.34 0.93 0.1830
Vision difficulties 35 0.16 35 0.38 1.20 35 0.20 35 3.04 2.19 0.0001 *
Accommodation difficulties 35 0.19 35 0.26 1.44 35 0.03 35 3.07 2.54 0.0001 *
Brow ache 35 0.06 35 0.14 0.65 35 0.20 35 2.40 2.72 0.0001 *
Bothersomeness of
Ocular symptoms 35 0.13 35 0.11 0.46 35 0.09 35 0.78 1.05 0.0001 *
Taste side effects 35 0.01 35 0.00 0.12 35 0.09 35 0.19 0.57 0.0636
Vision difficulties 35 0.10 35 0.17 0.78 35 0.06 35 1.93 1.64 0.0001 *
Accommodation difficulties 35 0.20 35 0.00 1.06 35 0.01 35 1.84 1.85 0.0001 *
Brow ache 35 0.11 35 0.00 0.54 35 0.09 35 1.60 1.88 0.0001 *
Limitations on
Driving 34 0.00 34 0.15 0.62 34 0.00 34 1.68 1.81 0.0001 *
Reading 34 0.06 34 0.12 0.72 35 0.00 34 1.61 1.77 0.0001 *
Moderate activities 35 0.00 35 0.01 0.06 33 0.00 33 0.49 1.26 0.0275*
Global
6. QOL interfered with by
side effects? 35 0.20 34 0.00 0.95 35 0.09 35 1.57 1.82 0.0001 *
9. QOL interfered with by
activity limitations? 35 0.03 35 0.11 0.72 35 0.03 35 1.34 1.64 0.0001 *
10. How often did you miss
one or more doses? 35 0.23 35 0.20 1.16 35 0.31 35 1.14 1.90 0.0144*
12. How satisfied have you
been with the test
medication? 35 0.97 35 0.26 1.54 35 0.57 35 2.29 1.84 0.0001 *

QOL = quality of life

* Significant at the 0.05 level of significance.

• Three limitations-of-living-activities domains (limi-
tations with driving, reading, and moderate activi-
ties).
As well, we retained four of the five prespecified global
questions: 6, 9, 10, and 12. This validation study has
shown that 11 ofthe 13 domains (all but the frequency-of-
taste and bothersomeness-of-taste domains) and 4 global
questions (6, 9, 10, and 12) of the COMTOL question-
naire have good-to-excellent internal consistency, relia-
bility, reproducibility, construct validity, discriminant va-
lidity, and responsiveness. Therefore, the COMTOL ques-
tionnaire has met validation criteria and is suitable for
use in clinical trials comparing topical ophthalmic medi-
cations.
The list of side effects in questions 4 and 5 was con-
'structed from common side effects of topical therapies for
lowering lOP, as reported by patients in clinical trials.
For a topical medication, these consist primarily of ocular
and other local effects, and effects on visual function.
Because systemic side effects of ,8-blockers occur infre-
quently in patients who do not have contraindications, 8 •9
we did not include systemic side effects in questions 4
and 5. In a study of 2,217 patients, Katz et al 8 discovered
that the most frequent systemic adverse experience among
patients treated with timolol was dizziness at 1%. In a
study of 489 patients, Wilson et al 9 discovered that the
most frequent systemic adverse experience of patients
treated with timolol was cardiovascular at 1.4%, followed
by headaches at 1.2%, and dizziness at 0.6%. Clearly in
the population included in the present trial, which ex-
cluded patients with contraindications to timolol or other
,8-blockers, systemic side effects would occur at a very
low rate.
Although these systemic side effects are not directly
queried in the questionnaire, any interference which such
systemic effects may have on living activities such as
reading, driving, or walking several blocks would be cap-
tured by the questionnaire. Including patients taking a ,8-
blocker (timolol) to help identify activities that might
be adversely affected by medication provides additional
evidence that the questionnaire addresses adverse effects
of ,8-blockers on living activities. The tolerability ques-
tionnaire is intended to supplement, but not to replace,
the reporting of spontaneously occurring adverse events
which is a part of any clinical trial. Using both together
may permit a more comprehensive assessment of the

341
 Ophthalmology Volume 104, Number 2, February 1997
safety and tolerability profiles of the ophthalmic agents
being evaluated in the trial.
Although some side effects listed in questions 4 and
5 did not occur frequently in this study with timolol
and pilocarpine, they may be common local or visual
side effects of other topical ophthalmic medications.
Therefore, we did not eliminate any side effects so that
the questionnaire could be used in comparative studies
of other topical ophthalmic medications used to treat
lOP.
Specifically, although there were very few reports of
bitter or unusual taste, we kept the frequency-of-taste and
the bothersomeness-of-taste domains because we know
that bitter or unusual taste is a side effect of dorzolamide
(a new topical CAl agent). However, this study could not
adequately evaluate the measurement characteristics of
these two domains.
During the validation process, global question 11 (why
eye drops were missed) was eliminated due to low re-
sponse and lack of ability to discriminate between groups.
As noted in the results, redundant questions (7 and 8) in
the list of routine activities were also eliminated.
Although the COMTOL questionnaire asks patients
about the impact of side effects on their quality of life
(global questions 6 and 9), the questionnaire is limited in
that it does not assess the impact on each individual do-
main of health-related quality of life. Thus, perhaps a
generic health-related quality-of-life questionnaire (e.g.,
the SF-36 10) could be administered along with the COM-
TOL in order to describe more completely the impact of
side effects on a full range of health-related quality-of-
life domains.
The COMTOL can also be used in studies comparing
antiglaucoma medication added to a topical ,8-blocker
used as first line therapy. However, at the end of such a
study, the conclusions that are drawn must be stated in
terms of one regimen (additive therapy plus first- line
therapy) being better tolerated than another, as opposed
to one additive therapy being better tolerated than an-
other. Also, the therapies (both additive and first-line)
should be such that their common side effects are among
those listed in questions 4 and 5.
A caution should be extended for any questionnaire
342
that attempts to discriminate between treatments based
on tolerability of common side effects. Before comparing
how common side effects of different medications affect
patients, the treatments must have been adequately stud-
ied so that the most common side effects of the different
medications are known, documented, and included in the
list of side effects that are examined by the questionnaire.
We believe that the COMTOL includes the most com-
mon ocular and other local side effects, and effects on
visual function, of topical ophthalmic agents for treating
lOP, and provides a method for assessing how such medi-
cations affect common living activities. Of course, tolera-
bility is highly patient-specific and results established in
clinical trials cannot be assumed to apply to patients with
other medical conditions that were excluded from the
trials and that may affect medication tolerability.
References
1. Jolliffe IT. Principal Component Analysis. New York:
Springer-Verlag; 1986:124-8.
2. Cronbach LJ. Coefficient alpha and the internal structure
of tests. Psychometrika 1951; 16:297-334.
3. Stewart AL, Ware JE. Measuring Functioning and Well-
being: The Medical Outcomes Study Approach. Durham,
NC: Duke University Press; 1992, pp81-5.
4. Deyo RA, Diehr P, Patrick DL. Reproducibility and health
status measures: statistics and strategies for evaluation.
Control Clin Trials 1991;12:142S-158S.
5. Streiner DL, Norman GR. Health Measurement Scales: A
Practical Guide to Their Development and Use. New York:
Oxford University Press; 1989, pp89-90.
6. Lin LK. A concordance correlation coefficient to evaluate
reproducibility. Biometrics 1989;45:255-68.
7. Mosteller F, Tukey JW. Data Analysis and Regression: A
Second Course in Statistics. Reading, MA: Addison-Wes-
ley Pub Co; 1977, pp145-8.
8. Katz IM, Kulaga SF Jr, Gould AL, et al. Long-term tolera-
bility and efficacy of timolol ophthalmic solution. Glau-
coma 1987;9:84-8.
9. Wilson RP, Spaeth GL, Poryzees E. The place of timolol
in the practice of ophthalmology. Ophthalmology 1980; 87:
451-4.
10. Ware JE Jr. SF-36 Health Survey Manual and Interpretation
Guide. Boston, MA: The Health Institute, New England
Medical Center, 1993.