Development and Validation of Disease-Specific Measures for Choroidal Melanoma: COMS-QOLS Report No. 2

Objective  To develop and validate scales measuring common concerns of patients with choroidal melanoma: perception of appearance, concern about cancer recurrence, and difficulty with vision-dependent activities requiring stereopsis or binocularity.

Design  Cross-sectional study within a randomized multicenter clinical trial.

Methods  Eight-hundred forty-two of 1317 patients with choroidal melanoma enrolled in the Collaborative Ocular Melanoma Study (COMS) for medium-sized tumors and randomized to receive iodine 125 brachytherapy or enucleation were interviewed. Scale reliability was evaluated using Cronbach's α, and validity was investigated through correlation with existing scales and with data collected during COMS clinical examinations.

Results  All 3 proposed scales had good internal consistency reliability. The appearance and recurrence scales had low to moderate correlation with the 36-Item Short-Form Health Survey mental health scale (r = 0.26 and 0.31, respectively) and with the the Hospital Anxiety and Depression Scale depression (r = −0.22 and−0.19) and anxiety (r = −0.27 and −0.42) scales. Appearance scores were significantly associated with occurrence of appearance-altering complications and conditions, recurrence scores were associated with recent removal of the brachytherapy-treated eye, and stereopsis/binocularity scores were higher in patients with good visual acuity in both eyes than in those with good visual acuity in one eye and an enucleated fellow eye.

Conclusions  All 3 proposed scales have good internal consistency, range, and SDs of measurement in the tested population. Based on clinical data, there is evidence of good construct validity for all 3 scales, although there also is evidence that the stereopsis/binocularity scale is sensitive to other aspects of vision in addition to stereopsis and binocularity. The appearance and recurrence scales capture clinically relevant information not available from standard mental health scales.

HEALTH- AND VISION-related quality of life assessment has become increasingly common in ophthalmologic studies, particularly in clinical trials.^1-9 Participants in a 1988 workshop charged with reviewing current issues and knowledge in the field of quality-of-life assessment and applicability to visual conditions noted that "demonstrating a way to improve the quality of life may be a key benefit of a trial, being just as important as such traditional outcomes measures as changes in clinical manifestations and increased longevity."¹⁰

The Collaborative Ocular Melanoma Study (COMS), a set of clinical trials sponsored by the National Eye Institute of the National Institutes of Health, included a randomized trial for choroidal melanoma whose primary goal was to determine whether treatment with eye-sparing brachytherapy or enucleation offered patients the best chance of survival.¹¹ This COMS trial enrolled eligible patients with medium-sized unilateral choroidal melanoma and randomly assigned them with equal probability to treatment with enucleation or brachytherapy using an iodine 125 plaque. Recruitment for the trial began January 22, 1987, and ended July 31, 1998, with 1317 patients enrolled. Patient follow-up continues until July 31, 2003. The primary outcome of the trial was overall survival; secondary outcomes included cancer-free survival and visual acuity.

The original design of the COMS did not incorporate measures of disease impact on quality of life, except for periodic measurement of visual acuity.^12,13 However, it was recognized that the 2 treatment approaches—enucleation and brachytherapy—were likely to have a different impact on patients with respect to visual functioning, body image, and concerns about cancer recurrence. These potential differences have become of particular importance given that the treatment and follow-up costs associated with brachytherapy are presumed to be greater than those for enucleation and that no survival difference between treatments has been found.¹⁴

An ancillary component of the COMS, the COMS Quality of Life Study (COMS-QOLS), was initiated in 1995 and was designed to measure the impact of choroidal melanoma on health- and vision-related quality of life and to determine whether health- and vision-related quality of life differed according to treatment. The COMS QOL interview consists of the 36-Item Short-Form Health Survey (SF-36), the Activities of Daily Vision Scale (ADVS), the National Eye Institute Visual Functioning Questionnaire (NEI VFQ), and the Hospital Anxiety and Depression Scale (HADS).^15-20 The rationale for choosing these questionnaires is described in a previous publication.¹³ During the development of the COMS-QOLS, focus group and pilot studies with non-COMS patients with choroidal melanoma identified additional concerns that were not specifically addressed in the generic or vision-specific questionnaires, namely, effect of treatment on appearance and concern about cancer recurrence. Both have been reported as major concerns of patients with other cancers.^21-30 However, at the time the COMS-QOLS was being designed, there was a lack of suitable scales that could be used to measure these concerns.^30,31 Also, although the ADVS and NEI VFQ measured level of difficulty with a wide range of vision-dependent activities, they did not include a scale intended to measure difficulty with tasks requiring stereopsis or binocularity, and the feasibility of using items from these scales to construct a scale reflecting difficulty with these tasks had not been investigated. All 3 of these measures were hypothesized to differ by treatment in COMS patients.

Specifically, COMS-QOLS investigators hypothesized that patients treated with brachytherapy, who retain the affected eye and usually retain useful vision initially, would be more satisfied with their appearance after treatment and have less difficulty performing tasks requiring stereopsis or binocularity. However, because the eye is retained, it was hypothesized that these patients would have greater anxiety and concern about recurrence of their cancer than patients whose eye, and presumably all tumor, has been removed. Differences by treatment on all 3 of these scales were hypothesized to diminish with time after treatment, as brachytherapy patients lost visual acuity owing to radiation retinopathy or cataract and the cancer diagnosis became less immediate and as enucleation patients adapted to seeing and performing visually oriented tasks with one eye and to possible changes in appearance. The purpose of this article is to describe the development of scales designed to measure these specific concerns of patients with choroidal melanoma and to establish that the scales are sufficiently reliable and valid that they may be used for assessment and comparison of choroidal melanoma patient populations in the COMS and other studies.

All patients were participants in the COMS multicenter clinical trial for patients with medium-sized choroidal melanoma and the COMS-QOLS, the ancillary study of quality of life. The designs of both studies have been described in detail elsewhere.^11,13

Interviews for the ancillary COMS-QOLS were initiated February 22, 1995, and initially included baseline, 6-month, and annual follow-up health-related quality of life interviews by telephone for newly enrolling COMS patients only. The study was expanded in 1997 to include annual interviews for patients enrolled in the COMS before 1995. The institutional review boards of all participating institutions reviewed and approved the protocols for the COMS Medium Tumor Trial and the COMS-QOLS. All patients participating in the COMS-QOLS gave written consent to enrollment in both studies. As of December 31, 2000, 842 of the 1317 patients enrolled in the COMS Medium Tumor Trial had given informed consent to participate in the ancillary COMS-QOLS. The remaining patients died before study initiation (n = 231), declined to participate (n = 214), or were being seen in one of a few clinical centers that elected not to participate in the ancillary study (n = 30). Two data sets were analyzed for this study: the first consisted of the most recent interview completed by enrolled patients as of December 31, 2000, and the second consisted of the first interview ever completed by enrolled patients that included the concern about recurrence items. Because the appearance items were added to the interview at a later date than the recurrence items, many patients did not answer the appearance items on the first interview. Hence, analyzing the most recent interview increases the sample size available for analysis. However, because of concerns that familiarity with the questions might have some effect on estimates of reliability, the data set based on the first interview also was analyzed. Results from the 2 analyses were similar, so only the results based on analysis of the most recent interview data are presented.

After a review of pertinent literature and existing questionnaires, 5 questions concerning patient perception of appearance and 3 questions concerning patient fears of cancer recurrence were drafted by the COMS-QOLS investigators in consultation with the COMS-QOLS Advisory Committee. The questions were pilot tested for content and acceptability in patients with choroidal melanoma at a large COMS clinical center (Bascom Palmer Eye Institute) before their refinement and incorporation into the COMS-QOLS interview (Table 1).

Two independent reviewers with clinical expertise in choroidal melanoma were asked to review items from the ADVS and the NEI VFQ and to identify all items relating to tasks requiring stereopsis or binocularity. These items, with one exception, were included in analyses as possible stereopsis/binocularity scale items. The exception was for items related to driving a car. These items were missing (not answered) more frequently than almost any other items from the ADVS and the NEI VFQ. In nearly all cases of missing values, the patient indicated that he or she was no longer driving for reasons unrelated to vision, probably a reflection of the relatively older age of the COMS patient population. Also, because some individual driving items were expected to be sensitive to stereopsis, the overall score for the driving scales should be sensitive to loss of stereopsis and binocularity at least to some degree. Hence, the driving items were not included in the stereopsis/binocularity scale. Another item, "seeing things off to the side, " was not related to stereopsis but was related to monocularity and was expected to differ by the 2 COMS treatments. This item was included in the stereopsis/binocularity scale. This item could be eliminated when stereopsis alone is the intended target concept.

Items were scored by assigning a score of 0 to the answer indicating the poorest outcome, 100 to the best outcome, and an intermediate value that was determined by dividing the 0 to 100 scale into equally spaced intervals for intermediate outcomes (Table 1).Scores for items within each of the hypothesized new scales (concern about recurrence, perception of appearance, and stereopsis/binocularity) were then averaged to create a score for each of the new scales. For all scales except the appearance scale, when a patient did not answer some items within a scale, the remaining items were averaged to compute the score for that scale, unless fewer than half the possible items were answered. In this case, the score for that scale was considered missing. For the appearance scale, because item means for the first 2 items differed significantly from those of the last 3 items, the items were subgrouped according to mean, and missing values were imputed by averaging the nonmissing values within the subgroup. For the stereopsis/binocularity, recurrence, and appearance scales, a score could not be computed for 16 patients(1.9%), 20 patients (2.4%), and 7 patients (0.9%), respectively, after imputation of missing values.

Each of the 2 data sets was first analyzed including only patients with complete data for all items in the scales and was subsequently analyzed including all possible patients using the procedure for imputing missing data described previously. Results for the 2 analyses were similar; results based on imputing missing data are reported.

Descriptive statistics were computed to investigate the properties of the newly compiled scales and to demonstrate that they had adequate variability for their intended use without excessive floor or ceiling effects (percentage of patients scoring at the minimum or maximum value) or missing data. Item analysis and multitrait scaling analysis were performed to evaluate the extent to which the scales were measuring distinct concepts.^32,33 This included computing correlation of the score for each item with the score for its hypothesized scale (excluding that item) and for other scales included in the COMS-QOL interview. Items that correlate highly with their hypothesized scale and do not correlate with other scales have the best ability to discriminate across scales. Cronbach's α was computed for each scale to establish internal consistency reliability.³⁴

Spearman correlation coefficients for the new scales vs the standard scales included in the COMS-QOL interview were computed to determine whether expected patterns of correlation could be demonstrated and whether the new scales contributed information not captured by the standard scales. Specifically, the correlations of recurrence and appearance scale items with the SF-36 mental health scale (MHI-5), the HADS anxiety and depression scales, the SF-36 physical functioning scale, the NEI VFQ general vision scale, and the overall ADVS were determined. The COMS-QOLS investigators hypothesized that the 2 new scales, and items from those scales, would have low to moderate correlation with the mental health scales and low or no correlation with the general health and visual functioning scales. The recurrence scale scores were compared for patients who had a diagnosis of another primary tumor or had their brachytherapy-treated eye removed during follow-up. It was hypothesized that these patients would have heightened concerns about recurrence of their choroidal melanoma and would have lower recurrence scores relative to other patients with choroidal melanoma. Spearman correlations between the stereopsis/binocularity scale and the other visual functioning scales were computed to determine the extent to which the new scale contributed information that was not captured by the other visual functioning scales.

To establish the construct validity for each of the new scales, we compared the median scale scores for those with and without the most relevant clinical variables using box plots with the Wilcoxon rank sum test or the Kruskal-Wallis test. For example, we expected that appearance scores would be lower for patients who were found to have ptosis or a pupillary defect during a COMS clinical examination.

Finally, patients for whom visual acuity data were available were divided into 4 groups for purposes of comparing and validating the stereopsis/binocularity scale score. Group 1 (n = 92) consisted of patients who had good visual acuity(20/40 or better) in both eyes. Most of these patients would be expected to have had good stereopsis. Group 2 (n = 342) consisted of patients with good visual acuity in one eye and an enucleated study eye. These patients had good visual acuity but had no stereopsis by definition. Group 3 (n = 223) consisted of patients who had good visual acuity in one eye and impaired visual acuity in the other (worse than 20/40). These patients probably had poorer stereopsis on average than patients in group 1. Group 4 (n = 14) consisted of patients with impaired visual acuity in both eyes or impaired visual acuity in one eye and an enucleated study eye. Most of these patients would be expected to have poor or no stereopsis. All group 4 patients had impaired visual acuity relative to the other 3 groups. It was hypothesized that group 1 would have the best stereopsis/binocularity scale scores, followed by group 3. Groups 2 and 4 were hypothesized to have the worst or lowest stereopsis/binocularity scale scores. Stereopsis/binocularity scale scores first were compared among the 4 groups, and then the z statistic from the Wilcoxon rank sum test comparing stereopsis score as a function of group 1 vs group 2 membership was computed and compared with the z statistics for the visual functioning scales from which stereopsis items were drawn. The stereopsis scale, if it was more sensitive than the other visual functioning scales to complete absence of stereopsis in conjunction with good visual acuity, would be expected to have higher z statistic values in this analysis than the other scales. The SF-36 physical functioning score also was included in the analysis as representative of a scale not expected to show sensitivity to loss of stereopsis and, therefore, to have a low z statistic value.

The patient population was approximately evenly divided between men and women; their median age at the time of last interview was 66 years (range, 26-91 years) (Table 2). As expected owing to the epidemiologic characteristics of choroidal melanoma, nearly all patients were non-Hispanic whites. Most patients (77%) had at least a high school education; educational level was unknown for some participants (4%). At the most recent interview, time since enrollment in the COMS and treatment for choroidal melanoma ranged from 0 to 13.3 years (median, 6.5 years). Visual acuity information was available for 671 (80%) of 842 patients at the time of the most recent visit. Patients who did not have visual acuity information generally had completed only a telephone interview and not a clinic visit at the time of the most recent scheduled follow-up. Most patients (84%) for whom information was available had good visual acuity (20/40 or better) in one eye, and the other eye had been removed (51%) or had visual acuity worse than 20/40 (33%). Few patients (14%) had good visual acuity in both eyes, and even fewer (2%) had visual acuity worse than 20/40 in both eyes or in the only remaining eye.

The descriptive statistics for the appearance and concern about recurrence scales, along with statistics for each of their component items, are given in Table 3. The mean (SD) score for the concern about recurrence scale was 74.5 (19.6), with 18% of patients scoring at the ceiling and 1% at the floor. The mean (SD) score for the perception of appearance scale was 71.1 (15.5), with 2% of patients scoring at the ceiling and 0% at the floor. For the appearance scale, scores for the first 2 component items (comparing the appearance of the treated eye with the other eye and with its appearance before treatment) are significantly lower than scores for the last 3 component items (feeling uncomfortable around others and satisfaction with eye and overall facial appearance). These first 2 items eliminate what would be a strong ceiling effect if the scale were based on the last 3 items only, as a high proportion of patients are very satisfied with their appearance; yet, some very satisfied patients agree that their eye appears somewhat or much worse when asked to compare with an objective standard such as their other eye. The median time for patients to complete the appearance and concern about recurrence scales and 4 demographic questions was 3.2 minutes (range, <1-20 minutes).

Individual recurrence and appearance items correlated well with their hypothesized scales (Spearman r = 0.63-0.90) and did not correlate well with other scales. For example, appearance and recurrence scale items had low to moderate correlations (r =−0.38-0.30) with the other mental health scales, as hypothesized, and no correlation with SF-36 physical function (r =−0.10-0.13), as expected. (For details, visit the COMS-QOLS Web site at http://www.jhu.edu/wctb/coms/qol.) Cronbach's α for the recurrence scale was .80 and for the appearance scale was .82. The NEI VFQ general vision and ADVS overall scales had low correlation with the appearance and recurrence scales, with patients who had poor vision functioning on either scale showing a modest tendency toward greater concerns about choroidal melanoma recurrence and their appearance.

Patients who underwent enucleation and who had less mobility of the prosthesis (P = .001), ptosis (P = .07), or a complication with the prosthesis, eyelids, or orbit(P<.001) had significantly lower appearance scale scores than those who did not have these problems (Figure 1). Complications in the eyes that received brachytherapy, that is, pain, dryness, and lack of orthophoria, generally were not significantly associated with appearance scale score, although only a few patients had these complications and there was a trend for lower appearance scale scores in patients who had them. Pupillary defect was commonly observed in eyes treated with brachytherapy and was significantly associated with lower appearance scale scores (P = .03) (Figure 1). Patients who had a second primary tumor diagnosed during follow-up did not differ from those without this diagnosis on the recurrence scale score (P = .63), but patients who had their brachytherapy-treated eye recently removed showed a trend for lower recurrence scale scores compared with brachytherapy-treated patients who still had their eye (P = .08) (Figure 2). However, no adjustment was made for multiple testing, and it is possible that some of the less significant P values may represent chance associations.

Neither the recurrence scale nor the appearance scale exhibited particularly strong correlations with the mental health scales included in the COMS-QOL interview. The strongest correlations were between the concern about recurrence vs HADS anxiety and MHI-5 scales (r = −0.42 and 0.31, respectively). Correlation with the HADS depression scale was weaker(r = −0.19). Correlations between the appearance scale and mental health scales also were relatively weak (r = −0.27 for HADS anxiety, r = −0.22 for HADS depression, and r = 0.26 for MHI-5).

Descriptive statistics for the stereopsis/binocularity scale and its component items are given in Table 4.Descriptive statistics for related but nonoverlapping visual functioning scales(ADVS glare and NEI VFQ role limitation, distance vision, and dependency scales) also are included in Table 4 for comparison purposes. The mean (SD) stereopsis/binocularity scale score was 78.1 (17.6), with 7% of patients at the ceiling and less than 1% at the floor. The SD of the proposed stereopsis/binocularity scale is comparable to that of other visual functioning scales. Furthermore, many fewer patients scored at the ceiling of this scale vs the scales used for comparison and vs nearly all the other ADVS and NEI VFQ scales. This finding is encouraging given that vision-oriented tasks that depend on stereopsis and binocularity are hypothesized to be most greatly affected by treatment for choroidal melanoma because most patients have good vision in at least one eye.

Table 5 gives the Spearman correlation of the stereopsis/binocularity scale with the other visual functioning scales. This scale had high correlation with the distance and near activities scores from the ADVS and the NEI VFQ and with the ADVS overall score and moderate correlation with other visual functioning scales. Multitrait scaling analysis of stereopsis scale items showed that all items also had moderate to high correlation with other visual functioning scales. In general, stereopsis items correlated better with the stereopsis scale than with the other visual functioning scales, although a few stereopsis items correlated slightly better with their original scale.

Comparing patients who were grouped according to levels of visual acuity and stereopsis, as expected, patients in group 1 (good stereopsis) had significantly higher stereopsis/binocularity scale scores than patients in all other groups(mixed levels of stereopsis or no stereopsis) (Figure 3). Also, based on a comparison of group 1 (good stereopsis and good visual acuity) and group 2 (no stereopsis and good visual acuity) using the Wilcoxon rank sum test, the stereopsis scale was more sensitive to the absence of stereopsis than the other visual functioning scales from which stereopsis items were drawn. (For details, visit the COMS-QOLS Web site at http://www.jhu.edu/wctb/coms/qol.) This was true even for scales that exhibited high correlation with the stereopsis/binocularity scale, for example, the ADVS overall, ADVS distance vision, NEI VFQ distance activities, and NEI VFQ near activities scales (Table 5). The exception was for the single-item NEI VFQ peripheral vision scale; however, as discussed previously, this item does not measure impact of loss of stereopsis on visual functioning but rather loss of visual field, which is confounded with loss of stereopsis in the COMS patient population because of monocularity.

Cronbach's α for the 10-item stereopsis/binocularity scale was.90. Deleting component items from the scale lowered Cronbach's α only slightly. Hence, if there was interest in creating a briefer scale, deleting 1 or more items from the stereopsis/binocularity scale could be considered. For example, properties of a stereopsis/binocularity scale composed of the subset of 4 items on the 10-item scale derived from the NEI VFQ were examined. This scale also had good properties vis-à-vis mean (74.7), SD (19.1), and percentage of patients at the ceiling (12%) and floor (<1%). Again, items correlated better with the hypothesized stereopsis/binocularity scale(r = 0.76-0.78) than with the comparison visual functioning scales (r = 0.39-0.61). Cronbach's α for this scale was .80. The 10-item scale was retained for use in the COMS-QOLS because the NEI VFQ was added while the study was ongoing; therefore, a stereopsis/binocularity scale score based solely on the NEI VFQ would be missing for all interviews performed early in the study.

All 3 of the proposed scales exceeded the criterion of internal consistency for group-level comparisons³⁵ and had range and SDs of measurement that were comparable to other previously validated scales included in the COMS-QOL interview. Although many of the individual items had significant ceiling effects, there was room on both scales for group means to increase or decrease. The similar SDs to those of the other scales, for example, those incorporated into the SF-36 and NEI VFQ, mean that changes over time of similar magnitude will be statistically detectable with the COMS-QOLS sample size. The same conclusion holds for differences in mean scores between treatment groups. For example, the COMS-QOLS was powered to detect at least 10-point differences between treatment groups during each of the first 5 years of follow-up for 6 of the 8 SF-36 scales. Thus, it also is powered to detect at least 10-point differences on the appearance, recurrence, and stereopsis scales. This size difference is comparable to the observed mean difference in appearance score between enucleated patients with good or fair mobility of the prosthesis (mean, 70.2) and poor mobility of the prosthesis (mean, 58.2), to the mean difference in recurrence score between patients who recently had their brachytherapy-treated eye removed (mean, 62.7) and patients who still have their brachytherapy-treated eye (mean, 74.8), and to the mean difference in stereopsis score between patients with good visual acuity in both eyes(mean, 86.6) and patients with good visual acuity in one eye whose other eye was enucleated (mean, 77.3).

Perception of appearance and concern about recurrence items had low to moderate correlation with HADS and SF-36 mental health scales, as expected, and seem to capture additional information not measured by those scales. Furthermore, appearance and recurrence scores were associated with clinical events in a way that supports that they are measuring the intended target concepts. The possible exception was the lack of association between the recurrence score and having a second primary tumor diagnosed; however, the recurrence items are specifically targeted to choroidal melanoma recurrence and, in retrospect, it was not surprising that they may not be sensitive to diagnosis of another tumor. Also, anxiety about cancer recurrence may be lower after the event has actually occurred and the patient has had an opportunity to adapt to this change.

There is evidence in this study that the stereopsis/binocularity scale is sensitive to differences in difficulty performing visually oriented tasks due to impaired stereopsis or lack of binocularity and is more sensitive than existing visual functioning scales. But other aspects of vision had some role as well. In particular, the difference on this scale between groups 1 (good visual acuity and good stereopsis) and 2 (good visual acuity and no stereopsis) shows that it is sensitive at the group level to the effect of good stereopsis vs no stereopsis in patients with good visual acuity. However, distribution of scores was nearly identical for groups 2 and 3 (good visual acuity and mixed level of stereopsis). The average level of stereopsis in group 3 was almost certainly better than that in group 2, suggesting that our sample size did not provide adequate power to differentiate between impaired and absent stereopsis or that the score may not be sensitive to impaired vs completely absent stereopsis or that the score is affected by other aspects of vision in addition to stereopsis. Group 4 (poor visual acuity and mixed level of stereopsis) had poorer scores than all the other groups, suggesting that visual acuity is one of the additional aspects of vision that has some effect on the stereopsis score.

A relationship similar to that seen for the stereopsis/binocularity scale between patients with good stereopsis and binocularity and those with no stereopsis or binocularity also was seen for several of the other visual functioning scales. Hence, there was evidence that some of the visual functioning scales already included in the ADVS and the NEI VFQ are sensitive to some degree to loss of stereopsis and binocularity. However, except for the ADVS night driving scale and the NEI VFQ driving and peripheral vision scales, all differences were of smaller magnitude than those seen for the stereopsis/binocularity scale. Further studies comparing a clinical measure of stereopsis with stereopsis/binocularity scale scores would be useful in determining whether the proposed scale and existing ADVS and NEI VFQ scales are sensitive to varying levels of stereopsis and not just to its presence or complete absence.

There are several likely explanations for these observed results. Vision is a complex neural and cognitive process. Visual acuity is an essential component of almost any visually oriented task and would be expected to affect responses to most items on visual functioning questionnaires. In a population of patients with varying levels of visual acuity, it may not be reasonable to expect to separate out particular questions as measuring problems or impairment due to reduced visual acuity vs lack of stereopsis or any other aspect of vision. Also, ability to perform any particular task may depend on several aspects of vision whose relative contributions may differ depending on the patient and his or her particular impairments. Given the range of responses seen among patients with similar levels and types of impairment, perceived difficulty performing visually oriented tasks, feelings of dependence on others, etc, are highly variable from patient to patient. Clinical measures such as visual acuity and visual field do not completely capture the individual's subjective experience of vision and the way that impairment affects his or her life. Thus, valid and reliable visual functioning questionnaires can provide useful information that augments clinical measures. As this study and others^36-39 have found, when determining how alternative treatments differentially affect patients' daily lives, it may be necessary to expand beyond standard questionnaires to account for differences in potential adverse effects and complications, both physical and mental.

Corresponding author and reprints: Michele Melia, Sc M, COMS Coordinating Center, Wilmer Clinical Trials and Biometry, Johns Hopkins University, 550 N Broadway, Ninth Floor, Baltimore, MD 21205-2010 (e-mail: mmelia@jhmi.edu).

Article

Submitted for publication August 29, 2002; final revision received February 18, 2003; accepted March 5, 2003.

This study was supported by the National Eye Institute through cooperative agreement EY10207 with the National Institutes of Health, Bethesda, Md.

Clinical Centers

The W.K. Kellogg Eye Center, Ann Arbor, Mich: Andrew K. Vine, MD (principal investigator), and Julie M. Willis, LPN, COMA (clinic coordinator). Emory Eye Center, Atlanta, Ga: Paul Sternberg, Jr, MD (principal investigator), and Jayne M. Brown, BA (clinic coordinator). Piedmont Hospital, Atlanta: James H. Frank, MD (principal investigator). The Wilmer Ophthalmological Institute, Baltimore, Md: Andrew P. Schachat, MD (principal investigator), and Warren T. Doll, Jr, COA; Mike Hartnett, COT; Ellen F. Greenberg, COT (1990-2001); and Marguerite Alexander, RN, BSN (1986-1995) (clinic coordinators). Retina Associates of Cleveland, Cleveland, Ohio: Z. Nicholas Zakov, MD (principal investigator), and Thomas A. Rice, MD (1987-1997); Kristen E. Dempsey; Geraldine Daley, CRA (1997-2000); and Susan E. Lichterman, RN(1991-1997) (clinic coordinators). Schepens Retina Associates, Boston, Mass: Clement Trempe, MD (principal investigator), and Zena F. Nies (clinic coordinator). Retina Study Center, Northwestern University Department of Ophthalmology, Chicago, Ill: Lee M. Jampol, MD (principal investigator), and Beth Chiapetta, RN; Lori Kaminski, BSN, MS, RN; Annie Munana, RN; Jill M. Koecher, BA; and Mimi Mansfield (1996-1997)(clinic coordinators). Cole Eye Institute, The Cleveland Clinic Foundation: Froncie A. Gutman, MD (principal investigator), and Susan E. Lichterman, RN; Laura J. Holody, COA; and Tina E. Kiss, COT, CCRA (1994-1997) (clinic coordinators). Ohio State University College of Medicine, Columbus: Frederick H. Davidorf, MD (principal investigator), and Cynthia S. Taylor, BS (clinic coordinator). Texas Retina Associates, Dallas: Dwain Fuller, MD (principal investigator), and Sally A. Arceneaux, COST, COA; Jean Arnwine; and Theresa Anderson, COA(1986-1997) (clinic coordinators). Atlantic Eye and Face Center, Cary, NC: Jonathan J. Dutton, MD (principal investigator), and Kim K. Peddle, COT (1994-2000), and Cynthia Lyons, AA (1994-1997) (clinic coordinators). Hermann Eye Center, Houston, Tex: Richard S. Ruiz, MD (principal investigator), and Dalia Vargas (clinic coordinator). The University of Iowa Hospital and Clinics, Iowa City: Thomas A. Weingeist, MD, PhD (principal investigator), and Connie Fountain, COT, CCRC, and Marcia Griffin (clinic coordinators). Doheny Eye Institute, University of Southern California School of Medicine, Los Angeles: A. Linn Murphree, MD (principal investigator), and A. Frances Walonker, CO, COMT, and Margaret Padilla (clinic coordinators). Jules Stein Eye Institute, Doris Stein Eye Research Center, Los Angeles: Robert E. Engstrom, Jr, MD, and Bradley R. Straatsma, MD (1987-1999) (principal investigators), and Robert Caldwell, PA; Robert D. Almanzor; and Karen Baranick, COA (1993-1995)(clinic coordinators). University of Wisconsin Hospital and Clinic, Department of Ophthalmology, Madison: Suresh R. Chandra, MD (principal investigator), and Kathryn F. Burke, BA; Jennie R. Perry, BS(1999-2000); Margo S. Blatz (1998-1999); and Betty Lewis (1987-1998) (clinic coordinators). The New York Eye Cancer Center, New York: Paul T. Finger, MD (principal investigator), and Jeannie L. Ryan, COT, LPN; Paula A. Muir (1997-1999); and Cynthia Pearlberg (1987-1996) (clinic coordinators). Bascom Palmer Eye Institute, Miami, Fla: Timothy G. Murray, MD (principal investigator), and Nicole Cicciarelli, CST; Joy M. Lo Bracco; and Tina S. Miracle, Ed S (clinic coordinators). Medical College of Wisconsin Eye Institute, Milwaukee: Judy E. Kim, MD, and William F. Mieler, MD (1986-1999) (principal investigators), and Troy Drescher, BS; Sharon A. Rekow (1988-2000); and Mary M. Andrus, BS(1999) (clinic coordinators). Cornell University, New York: David H. Abramson, MD (principal investigator), and Diane M. Murphy; Indira S. Rollins, COA, CST; and Camille A. Servodidio, RN, MPH, CRNO (1986-1997)(clinic coordinators). Casey Eye Institute, Oregon Health and Science University, Portland: David J. Wilson, MD (principal investigator), and Shirley Ira, COT (clinic coordinator). Mayo Foundation, Department of Ophthalmology, Rochester, Minn: Dennis M. Robertson, MD (principal investigator), and Margaret M. Ruszczyk, CCRA, and Mary Schumann, COMT (clinic coordinators). Associated Retinal Consultants, PC, Royal Oak, Mich: Antonio Capone, Jr, MD, and Raymond R. Margherio, MD (1986-2000) (principal investigators), and Kristi Cumming, RN, MSN; Patricia Manatrey, RN; and Beth Mitchell, RN (1989-1998) (clinic coordinators). The University of Texas Health Science Center, Department of Ophthalmology, San Antonio:Wichard van Heuven, MD (principal investigator), and Leticia S. Leija, RDMS, and Carolyn Torres, LVN (1993-1996) (clinic coordinators). Princess Margaret Hospital, Ocular Oncology Service, Toronto, Ontario: E. Rand Simpson, MD (principal investigator), and Lee Penney, COA, and Hugh McGowan, MD (clinic coordinators). Northern California COMS Group, San Francisco: Robert N. Johnson, MD (principal investigator), and Karen Laughlin, LVN; Margaret M. Stolarczuk, OD; Nubia E. Garcia, RN; and Renee Y. Guertin (1997) (clinic coordinators). Retina-Vitreous Consultants, Pittsburgh, Pa: Karl R. Olsen, MD, and Jeffrey S. Rinkoff, MD (1988-1996) (principal investigators), and Donna L. Green, RN, CRNO, and Barbara Bahr, LPN (1988-1999) (clinic coordinators). University of Virginia Health Sciences Center, Department of Ophthalmology, Charlottesville: Brian P. Conway, MD (principal investigator), and Mary J. (Jonni) Thoma, RN, BSN (clinic coordinator). Porter Adventist Hospital, Denver, Colo: Kenneth R. Hovland, MD (principal investigator), and Suzette Compton, RN (clinic coordinator). Midwest Eye Institute, Indianapolis, Ind: John T. Minturn, MD (principal investigator), and Donna J. Agugliaro, RN, BSN (clinic coordinator). Ophthalmic Consultants Northwest, Seattle, Wash: Edward B. McLean, MD (principal investigator), and Sheila M. Davidson (clinic coordinator). University of Washington, Seattle: Craig G. Wells, MD (principal investigator), and Betty Lawrence, BS, COA (clinic coordinator). Retina Associates of Florida PA, Tampa:W. Sanderson Grizzard, MD (principal investigator), and Janet Traynom, COT; William J. Malenfant (1996-1999); and Theresa L. Shannon, COA (1994-1995) (clinic coordinators). Retina Associates Southwest, Tucson, Ariz: Leonard Joffee, MD, FRCS(principal investigator), and Sally K. Brandon, LPN, COT, and Mari L. Bunnell, COA (clinic coordinators). Montreal Hospital, Montreal, Quebec: Christine Corriveau, MD (principal investigator), and Liette Bousquet and Suzanne Casavant, RN (clinic coordinators). Dean A. McGee Eye Institute, Oklahoma City, Okla: Reagan H. Bradford, Jr, MD (principal investigator), and Jason D. Jobson; Lisa M. Ogilbee, COA, BS; Jay Johnson, LPN (1999-2000); and Janie M. Shofner, COA, CCRA (1990-1998)(clinic coordinators).

Advisory Committee

B. Michele Melia, Sc M, and Claudia S. Moy, PhD (1994-2001) (chairpersons); David Cella, PhD; Beth Chiapetta, RN (1999); Li Ming Dong, PhD; John D. Earle, MD; Beth Erickson, MD (1995-1996); Connie Fountain, COT; James A. Hayman, MD; Kenneth R. Hovland, MD; Tina Kiss, COT (1995-1996); Natalie Kurinij, PhD; Carol M. Mangione, MD, MSPH; Kelly S. Manos, MAS (1995-1998); Lee McCaffrey, MA; Paivi Miskala, MSPH (1999); Beth Mitchell, RN (1997-1998); Timothy G. Murray, MD; Seth Reiner, MD (1997-1998); Sandra M. Reynolds, MA; Jeffrey Rinkoff, MD (1995-1996); Dawn Smith; and Michael Stefanek, PhD (1995).

Coordinating Center

Principal investigators: B. Michele Melia, Sc M, and Claudia S. Moy, PhD (1994-2001). Coinvestigator: Lee McCaffrey, MA. Statistician: Li Ming Dong, PhD. Study Coordinators: Dawn Smith; Kelly S. Manos, MAS (1994-1998); and Paivi Miskala, MSPH (1999). Systems Analyst: M. Marvin Newhouse. Programmers: Sandra M. Reynolds, MA; and Jonathan S. Kerman (1994-1998). Administrative Assistant: Lisa A. Lassiter. Interviewers: Robert G. Casper, MS; Alice D. Keith; Christine B. Alden; and Nancy Prusakowski, MS (1998-2001).

Data and Safety Monitoring Committee as of December 2001

Chairperson: Matthew D. Davis, MD. Voting members: Marc S. Ernstoff, MD, FACP; Frederick L. Ferris III, MD; Lael C. Gatewood, PhD; Robert J. Levine, MD; Robert W. Makuch, PhD; James Marks, MD; Travis Meredith, MD; and Donald L. Patrick, PhD, MSPH. Nonvoting ex officio members: Marie Diener-West, PhD; John D. Earle, MD; Stuart L. Fine, MD; Barbara S. Hawkins, PhD; Natalie Kurinij, PhD; and B. Michele Melia, Sc M.