Development of a vision-related quality of life instrument for children ages 8–18 years for use in juvenile idiopathic arthritis–associated uveitis

Authors


Abstract

Objective

To determine the validity and reliability of a novel questionnaire to measure vision-related quality of life (VRQOL) in children ages 8–18 years for use in juvenile idiopathic arthritis (JIA)–associated uveitis: the Effects of Youngsters' Eyesight on Quality of Life (EYE-Q).

Methods

Several steps validated the EYE-Q. We interviewed experts and children on how vision affects a child's activities. We developed new items and selected relevant items from existing instruments. We administered initial versions of the EYE-Q to normal-sighted children and those with JIA-associated uveitis. For this study, children with various (or no) ocular conditions were recruited from a clinical population. Visual acuity and contrast sensitivity were performed, and the EYE-Q and Pediatric Quality of Life Inventory (PedsQL) were administered. The EYE-Q was repeated 10 days later. Patients, parents, and physicians rated vision severity.

Results

Of 120 patients, 48% were female, 46.7% had no visual impairment, and 53.3% had bilateral eye involvement. The mean age was 11.3 years. There were significant differences in the measures based on visual acuity (P < 0.001). Children with more severe visual acuity and bilateral eye involvement had worse EYE-Q scores (P < 0.001). There were significant associations between the EYE-Q and PedsQL (r = 0.375), repeat EYE-Q (r = 0.864), and clinical measures of ocular disease (r = −0.620).

Conclusions

Our study provides evidence of the validity and reliability of the EYE-Q in the measurement of VRQOL. The EYE-Q may complement clinical measures of visual impairment and overall QOL and become an important tool in the assessment of QOL in JIA-associated uveitis.

INTRODUCTION

Visual impairment (VI) in children is a significant problem in the US. Approximately 6.8% of children ages <18 years have a diagnosed vision problem, and 25 per 1,000 children in the US experience VI or blindness (1, 2). Juvenile idiopathic arthritis (JIA)–associated uveitis is commonly associated with VI. JIA-associated uveitis is the most prevalent extraarticular manifestation of JIA in North America and occurs in 10–20% of children with JIA (3, 4). It is a chronic nongranulomatous inflammatory eye disease that accounts for up to 80% of all pediatric anterior uveitis (3). JIA-associated uveitis is associated with ocular diseases that can result in VI, such as cataracts, glaucoma, and cystoid macular edema (5–9).

Quality of life (QOL) is known to be an important construct, particularly related to chronic disease. Vision-related QOL (VRQOL) is related to, but not identical to, visual function. VRQOL represents the degree to which vision impacts an individual's ability to complete activities of daily living and one's social, emotional, and economic well-being. VRQOL can be assessed by measuring the degree of impairment experienced in activities of daily living that rely on sight (i.e., impaired daily function secondary to visual difficulties is a proxy for visual function). Therefore, significant VI would lead to difficulty in performing tasks that rely on vision.

A child's activities in the home and school largely depend on vision. Since uveitis sequelae can result in compromise of visual function, studies of children with uveitis should account for visual disability. Unfortunately, there are few validated instruments to assess VRQOL in children ages 8–18 years. Instead, we rely on the clinical ophthalmologic examination (e.g., visual acuity [VA], visual field, and contrast sensitivity [CS]), measures of overall QOL, or adult-based instruments (10–12). It is unclear whether these measures accurately represent the child's perspective of the impact of VI on their daily life. Importantly, studies in adult patients use standardized visual function questionnaires in addition to clinical measures to assess visual disability (National Eye Institute Visual Function Questionnaire [NEI-VFQ]) (13, 14).

To date, studies examining QOL and outcomes in JIA focus on physical function assessments (i.e., Childhood Health Assessment Questionnaire, joint counts) either as proxies for or predictors of QOL (15, 16). Therefore, current methods to assess QOL in JIA may underestimate the impact of the disease on a child's experience because they fail to account for disease-specific limitations, particularly those relating to vision. This led to the development of a new instrument to evaluate the performance of activities that rely on vision in the home and school: the Effects of Youngsters' Eyesight on Quality of Life (EYE-Q) (17). Our objective was to provide further evidence of the validity and reliability of our instrument for children ages 8–18 years.

Significance & Innovations

  • Studies that examine quality of life (QOL) and outcomes in juvenile idiopathic arthritis (JIA) focus on physical function assessments. This may underestimate QOL, as its measurement focuses on musculoskeletal function secondary to arthritis, and not the impact of uveitis and the contribution of visual disability to QOL.

  • There is a paucity of instruments that measure visual disability and vision-related QOL (VRQOL) in children. There are none that are specific for children with uveitis that would make the assessment of QOL related to visual disability difficult in this population. The Effects of Youngsters' Eyesight on Quality of Life (EYE-Q) could serve as a validated instrument that measures VRQOL in children with uveitis.

  • In JIA-associated uveitis, clinical measures are used to evaluate outcomes, hence the need for a more comprehensive evaluation that takes into account the child's perspective. The EYE-Q could complement ophthalmologic examinations and global health measures and augment the assessment of the impact and utility of treatment.

SUBJECTS AND METHODS

This was a validation study that was approved by the Institutional Review Board of Emory University and conformed to the requirements of the US Health Insurance Portability and Privacy Act. Consent was obtained from the parent/legal guardian, and assent was obtained from the children prior to administration of the instruments.

Phase 1: item generation.

We interviewed pediatric rheumatologists, pediatric ophthalmology professionals (e.g., ophthalmologists, optometrists, clinical research technicians), and children with and without VI regarding the ways vision affects a child's daily activities. We also selected relevant items from existing instruments and adapted them to increase their relevance to a US-based sample of children (18, 19). This led to the development of our initial instrument consisting of 20 items thought to measure VRQOL. Based on expert consensus, we therefore established face and content validity demonstrating that the items generated were appropriate and fully measured the complexity of the target construct (20).

Phase 2: operationalism.

Thirteen normal-sighted 8–18-year-old children then evaluated the questionnaire and items for comprehensibility, format, and relevance. Items were retained if ≥70% of respondents agreed that they performed the task and it was important to be able to do so. We also asked for feedback regarding additional tasks that rely on vision. This led to further refinement of the EYE-Q, and 13 items were retained as a result of this analysis.

Phase 3: pretesting.

The refined version of the EYE-Q was administered in a cross-sectional study of children with JIA with and without uveitis (21). Children with uveitis had worse EYE-Q scores compared to children without uveitis, supporting the utility of the EYE-Q as an assessment tool for VRQOL in children with JIA and uveitis. This establishes criterion- and construct-related validity by demonstrating that the EYE-Q correlates with external standards, criteria, or diagnoses, and conforms to the theoretical concepts or constructs it was designed to measure (20, 22).

Revised version.

After discussion with other pediatric rheumatology, ophthalmology, and epidemiology experts, additional questions for children ages 16–18 years were added. As children mature and engage in new developmental tasks, some of which are vision related, the ability or inability to perform certain tasks would likely contribute to QOL. Therefore, we included items related to driving from the 25-item NEI-VFQ (14). We also modified the phrasing of items for increased comprehensibility, and incorporated questions to evaluate photosensitivity, night vision, use of visual aids, and the patient's perception of overall visual function. This led to the current version of the EYE-Q used in this study.

Subjects.

A convenience sample of 120 children with and without ocular disorders was enrolled from the pediatric ophthalmology and uveitis outpatient clinics at the Emory Eye Center. All eligible children and their parents were approached while waiting for their ophthalmology appointment. Inclusion criteria included: 1) age 8–18 years and 2) at least a third-grade reading and comprehension level. Exclusion criteria included: 1) recently dilated, 2) inability to complete the questionnaires for any reason, 3) major developmental disorders, or 4) refusal to participate.

Demographic/baseline.

Parent- and patient-based questionnaires were completed in person. A member of the research team (STA-H, LP) was available to read the questions to subjects whose VI prevented completion of the instruments on their own. The individual administering the questionnaires was masked to the child's ophthalmologic diagnosis and best-corrected VA at the time of the visit. Date of visit, date of birth, medical record number, address, phone number, sex, ethnicity, race, and presence of ophthalmologic disease were collected from the parent/guardian. Medical record review confirmed eye disease and laterality of eye involvement.

Assessment of VRQOL.

The EYE-Q was administered to all child subjects who completed the age-appropriate modules accordingly. The EYE-Q is a patient-based self-report that consists of 23 items for children ages 8–15 years, and 26 items (3 additional items on driving) for children ages 16–18 years. Items consider tasks that rely on distance (6 items), near (7 items), color (1 item), and night vision (1 item); functionality (3 items); and photosensitivity (2 items). There is a subjective assessment of vision severity where the child rates his or her eyesight as: 1 (excellent), 2 (good), 3 (fair), 4 (poor), 5 (very poor), or 6 (blind). A 5-point Likert scale response format is used to assess difficulty performing each task, with response options including: 1 (not hard), 2 (a little hard), 3 (hard), 4 (very hard), and 5 (cannot do). Each item also provides the child with the option of endorsing “does not apply.” We chose not to use a visual analog scale, as this may be problematic for children with visual disability. The EYE-Q takes less than 10 minutes to complete.

The module for 16–18-year-olds contains items related to driving. If the teen is not driving or has never driven, the module is identical to that for 8–15-year-olds. If driving was given up because of eyesight, 2 points are deducted (similar to needing a visual aid). If the teen is driving presently, 3 additional items inquire about difficulty driving 1) during the day, 2) at night, and 3) in difficult conditions.

One question asks about the need for visual aids (special lamps, magnifying glass, and large-print material), and allows the child to specify other aids used. Two points are deducted for the use of each aid, with a maximum of 8 points (4 aids) deducted. An open-ended question asks for feedback regarding any other tasks that rely on vision.

The instrument is then scored as the sum of the items (minus the aids, if used) divided by the number of items answered (20 or 23 items, depending on the module). Items on vision assessment and additional VRQOL tasks were not included in the scoring. Items were rescaled; therefore, scores range from 0–4, with higher scores indicating better VRQOL. Scores would not be computed if the children did not complete, missed, or marked “does not apply” for more than 50% of the items for any reason.

Distance VA was measured using a Snellen chart and CS using the Pelli-Robson CS test. Best-corrected VA was converted to Logmar for statistical purposes. This provides data to determine construct validity under the assumption that those with worse VA will have poorer VRQOL.

Children were grouped by VI according to best-corrected VA: 1) normal vision: VA 20/20 to <20/40, 2) mild VI: 20/40 to 20/60, 3) moderate VI: 20/70 to <20/200, and 4) severe VI: 20/200 or worse. They were also grouped according to laterality: 1) bilateral normal vision: VA 20/20 to <20/40 in both eyes (OU); 2) bilateral mild VI: 20/40 to 20/60 OU; 3) bilateral moderate VI: 20/70 to <20/200 OU; 4) bilateral severe VI: 20/200 or worse OU; 5) unilateral mild VI: 20/40 to 20/60 in the worse eye, <20/40 in the better eye; 6) unilateral moderate VI: 20/70 to <20/200 in the worse eye, <20/40 in the better eye; and 7) unilateral severe VI: >20/200 in the worse eye, <20/40 in the better eye. Only 116 children were grouped by laterality, since 4 were unclassifiable: 1) 20/50 and 20/125, 2) 20/50 and 20/200, 3) 20/80 and no light perception, and 4) 20/80 and 20/200.

To test retest reliability or consistency of the measure, the EYE-Q was completed twice: first in the clinic and then at home 10 days after the initial assessment. Subjects were called as a reminder to complete the questionnaire and return it in a self-addressed stamped envelope provided by the investigators.

Parents and physicians completed assessments of vision severity and were asked to score how much they thought the child's life was being affected by their vision from 1–10, where 1 = vision does not affect how he/she functions in daily activities and 10 = vision affects how he/she functions in daily activities.

QOL assessment.

The Pediatric Quality of Life Inventory, version 4.0 (PedsQL), is a validated measure of health-related QOL in children and adolescents ages 2–18 years (23). Four core scales are measured on a 5-point-scale: 1) physical functioning (8 items), 2) emotional functioning (5 items), 3) social functioning (5 items), and 4) school functioning (5 items). Scores range from 0–100, with higher scores indicating better QOL. Aside from the overall QOL score (PedsQL total), there are physical health summary (PedsQL physical) and psychosocial health summary (PedsQL psychosocial) scores. For this study, all scores were calculated and child-report forms were administered.

Statistical analysis.

Data were analyzed using SPSS, version 8. Descriptive analyses of baseline demographic and clinical variables based on VI and laterality were performed. Continuous variables that met the assumption of normality were compared using the independent t-test or analysis of variance. P values less than 0.05 were considered significant. Associations between measures of VRQOL, overall QOL, and the ophthalmologic examination were determined using Pearson's correlation coefficient. Correlations <0.5 were weak, ≥0.5 to ≤0.7 were moderate, and >0.7 were strong. To determine test–retest reliability, we examined correlations between the EYE-Q scores at the first visit and 10 days after. Cronbach's alpha was performed for internal consistency of the items.

RESULTS

There were 120 children seen in the pediatric ophthalmology clinics who agreed to participate, and 60.8% returned the second EYE-Q after 10 days.

Demographics and disease characteristics.

The mean ± SD age was 11.32 ± 2.7 years, 11.7% were ages 16–18 years, 47.5% were female, 6.7% were Hispanic, and 60.8% were white (Table 1).

Table 1. Demographic and disease characteristics (n = 120)*
 Value
  • *

    Values are the number (percentage) unless otherwise indicated.

Demographics 
 Age, mean ± SD years11.3 ± 2.7
  8–15 years old106 (88.3)
  16–18 years old14 (11.7)
 Hispanic race (n = 112)8 (6.7)
 Ethnicity (n = 115) 
  White73 (60.8)
  African American38 (31.7)
 Male63 (52.5)
Disease characteristics 
 Ocular disease or condition107 (89.2)
 Duration of eye disease among those with ocular disease, mean ± SD years5 ± 0.6
 Patients requiring use of ≥1 visual aids25 (20.8)
 Laterality, visual acuity 
  Bilateral normal: 20/30 or better56 (46.7)
  Bilateral mild: 20/40 to 20/606 (5)
  Bilateral moderate: 20/70 to <20/2008 (6.7)
  Bilateral severe: ≥20/2007 (5.8)
  Unilateral mild: 20/40 to 20/60 in the worse eye, ≤20/30 in the better eye19 (15.8)
  Unilateral moderate: 20/70 to <20/200 in the worse eye, ≤20/30 in the better eye8 (6.7)
  Unilateral severe: ≥20/200 in the worse eye, ≤20/30 in the better eye12 (10)
  Other4 (3.3)

Although 89.2% of children had 1 or more ocular diseases or conditions, 46.7% had normal bilateral VA and 53.3% had varying degrees of VI (Table 1). Only 4 subjects had strictly refractive disorders, 9 had disorders affecting the uvea (i.e., uveitis, aniridia), 4 had disorders affecting the cornea (i.e., cornea transplant, abrasions), 19 had disorders affecting the lens (i.e., cataracts, aphakia), 10 had disorders affecting the retina (i.e., retinopathy of prematurity, oculocutaneous albinism, retinoblastoma), and 93 had other diseases (i.e., optic nerve atrophy, esotropia, strabismus, etc.). Visual aids were required by 20.8% (18% large-print material, 14.2% magnifying glass, 2.5% special lamps, and 6.7% others). Other aids included canes, binoculars, Braille materials, special computer applications, and reading rulers. A total of 30.1% had other nonocular diseases (e.g., JIA, lymphoma, sickle cell disease).

The mean EYE-Q score was 3.45 (range 0.8–4.0) and the mean PedsQL total score was 78.99 (range 14–100). The mean Logmar VA of the better eye was 0.44, which is equivalent to a VA of 20/50 to 20/60 and a mean CS of 1.72.

Validity and reliability of the EYE-Q.

The EYE-Q appears to measure a single major construct. Cronbach's alpha on the first EYE-Q was 0.903 and on the second EYE-Q was 0.925. When considering the age-specific modules, α = 0.870 for the 8–15-year-old module and α = 0.962 for the 16–18-year-old module. This provides evidence for strong internal consistency.

When we removed 4 items with high inter-item correlations in the EYE-Q, our Cronbach's α = 0.879, which provides evidence that a 16-item module for this population retained strong internal consistency.

The EYE-Q also demonstrates strong evidence for reliability. There was a strong correlation between the scores on the first and second administration of the EYE-Q (r = 0.864).

To evaluate construct- and criterion-related validity, variables potentially associated with VRQOL were assessed: the clinical ocular examination (VA and CS) and the PedsQL (Table 2). The EYE-Q was weakly associated with the PedsQL total, psychosocial, and physical scores (r = 0.375, r = 0.344, and r = 0.341, respectively), and moderately associated with VA of the better eye (r = −0.620) and CS (r = 0.565). The PedsQL was not related to VA (r = −0.049) or CS (r = 0.065).

Table 2. Correlations between measures of validity and reliability*
 RP
  • *

    EYE-Q = Effects of Youngsters' Eyesight on Quality of Life; PedsQL = Pediatric Quality of Life Inventory; PedsQL physical = PedsQL physical health summary score; PedsQL psychosocial = PedsQL psychosocial health summary score; VA = visual acuity.

  • Higher scores indicate better quality of life.

  • Using Pearson's coefficient.

  • §

    Higher values indicate worse vision and VA.

  • Lower numbers indicate better vision.

EYE-Q score (range of scores 0–4)  
 EYE-Q repeat (range of scores 0–4)0.864< 0.001
 PedsQL total (range 0–100)0.375< 0.001
 PedsQL physical (range 0–100)0.341< 0.001
 PedsQL psychosocial (range 0–100)0.344< 0.001
 Logmar far best-corrected VA, better eye (range −0.30 to 3.00)§−0.620< 0.001
 Contrast sensitivity, both0.565< 0.001
PedsQL total score (range 0–100)  
 Logmar VA, better eye (range −0.30 to 3.00)§−0.0490.593
 Contrast sensitivity, both0.0650.571

Visual impairment and QOL.

Mean scores of the EYE-Q, PedsQL, and vision assessments in children with and without any VI are shown in Table 3. There were significant differences in the EYE-Q scores, PedsQL total scores, and patient and physician vision assessments in children with normal vision (bilateral VA 20/20 to <20/40) compared to those with VI (any eye with a VA 20/40 or worse). There was no significant difference in parent vision assessments in both groups.

Table 3. Differences in scores of measures based on vision impairment*
 Normal vision (n = 56)Vision impairment (n = 64)P
  • *

    Values are the difference ± SD. EYE-Q = Effects of Youngsters' Eyesight on Quality of Life; PedsQL = Pediatric Quality of Life Inventory.

  • Normal vision: both eyes with a visual acuity (VA) of <20/40.

  • Vision impairment: any eye with a VA of 20/40 or worse.

  • §

    Higher scores indicate better quality of life.

  • Using t-test (<0.05).

  • #

    Higher values indicate worse vision severity.

EYE-Q (range 0–4)§3.71 ± 0.303.22 ± 0.74< 0.001
PedsQL total (range 0–100)§82.4 ± 12.376.0 ± 16.50.044
Patient global assessment (range 1–6)#1.7 ± 0.72.3 ± 1.20.003
Parent global assessment (range 1–10)#3.9 ± 3.05.6 ± 2.90.803
Physician global assessment (range 1–10)#2.1 ± 1.54.0 ± 2.5< 0.001

As expected, the mean EYE-Q score was significantly worse (P < 0.001) in children who used visual aids than in those who did not. The PedsQL total score (P = 0.140) was not significantly associated with the use of aids, although a significant difference in the PedsQL physical score (P = 0.042) was noted. This association with the PedsQL physical score may be secondary to the need for aids for daily functioning such as a cane or large-print material.

Mean scores of the EYE-Q and PedsQL categorized by degree of VI are shown in Table 4. As VI worsened, the children had significantly worse VRQOL scores (P < 0.001). Likewise, worse EYE-Q scores were seen in children with worsening eye involvement based on laterality (P < 0.001). Similar findings were noted on the second administration of the EYE-Q. There was no significant difference in the PedsQL based on laterality or VI.

Table 4. Association between quality of life measures (mean ± SD) and vision impairment based on visual acuity and laterality*
 No. (%)EYE-Q (range 0–4)PedsQL total (range 0–100)
  • *

    EYE-Q = Effects of Youngsters' Eyesight on Quality of Life; PedsQL = Pediatric Quality of Life Inventory; VA = visual acuity; VI = vision impairment.

  • Correlation between VA in better eye and EYE-Q (P < 0.001) and PedsQL (P = 0.425 using analysis of variance [ANOVA]).

  • Correlation between laterality of VI and EYE-Q (P < 0.001) and PedsQL (P = 0.286 using ANOVA).

Best-corrected VA in the better eye   
 Normal: 20/20 to <20/4094 (78.3)3.65 ± 0.3780.0 ± 14.4
 Mild VI: 20/40 to 20/6011 (9.2)2.99 ± 0.6272.3 ± 18.5
 Moderate VI: 20/70 to <20/2008 (6.7)2.83 ± 0.4878.5 ± 18.9
 Severe VI: 20/200 or worse7 (5.8)2.16 ± 1.1476.5 ± 12.0
 All subjects120 (100)3.45 ± 0.6379.0 ± 15.0
Laterality of vision impairment   
 Bilateral normal: 20/20 to 20/4056 (46.7)3.71 ± 0.3082.4 ± 12.3
 Bilateral mild VI: 20/40 to 20/606 (5.0)2.96 ± 0.6875.5 ± 17.7
 Bilateral moderate VI: 20/70 to <20/2008 (6.7)2.69 ± 0.4573.2 ± 22.4
 Bilateral severe VI: 20/200 or worse7 (5.8)2.16 ± 1.1476.5 ± 12.0
 Unilateral mild VI: 20/40 to 20/60 in the worse eye, <20/40 in the better eye19 (15.8)3.52 ± 0.5676.7 ± 20.2
 Unilateral moderate VI: 20/70 to <20/200 in the worse eye, <20/40 in the better eye8 (6.7)3.72 ± 0.3371.0 ± 11.7
 Unilateral severe VI: >20/200 in the worse eye, <20/40 in the better eye12 (10.0)3.58 ± 0.3380.9 ± 11.9

DISCUSSION

VI can have a negative impact on a child's daily life in relation to school, home, and social activities. VI may also lead to poor QOL in children with JIA-associated uveitis, since adults with uveitis have been noted to experience poor QOL (10, 13, 24, 25). Although 65% of children with JIA-associated uveitis are asymptomatic in early disease, up to 45% have ocular complications at their initial ophthalmology examination (7). A preliminary study reports the use of the EYE-Q in children with JIA-associated uveitis as a valid measure of VRQOL and demonstrates that VI is an important component of QOL in children with uveitis (17). Data from our study suggest that our EYE-Q has construct- and criterion-related validity and test–retest reliability.

VI is expected to worsen an individual's QOL. Using the adult-based Low Vision Quality of Life Questionnaire, children with VI had significantly lower QOL scores than children with normal eyes (12). The EYE-Q was able to detect differences in children with VI as measured by VA. Of even greater importance, VRQOL appeared to worsen as VI became more severe, since EYE-Q scores decreased accordingly (lower scores = worse outcome). This provides evidence for construct validity, since the EYE-Q identifies differences in normal and impaired eyes and differences in the severity of VI.

Bilateral eye disease results in poorer VRQOL, which is supported by data from our study. Lower EYE-Q scores were significantly associated with bilateral eye disease and distinguished the severity of bilateral involvement. Although there were significant differences in the EYE-Q scores (P < 0.001), children with unilateral eye involvement did not appear to have worsening EYE-Q scores compared to children with bilateral involvement. Intuitively, one would expect that having one normal eye should adequately compensate for a visually impaired eye and allow a child to participate in normal activities without difficulty. This was demonstrated in our study. Despite severe impairment in the worse eye (20/200 or worse), as long as the better eye was normal, VRQOL and overall QOL scores did not appear to differ in the unilateral groups. This provides further support for construct validity of the EYE-Q.

A study by Wong et al showed that healthy adolescents with VI had statistically significant but not clinically impaired health-related QOL as measured by the PedsQL, and that refractive errors did not have an impact on overall QOL (11). In our study, the EYE-Q had significant mild correlations with the PedsQL and moderate correlations with VA and CS. The PedsQL was not significantly associated with VA. This indicates that although the PedsQL is a valid and reliable measure of overall QOL (23), it is a relatively poor measure of QOL affected by vision, and thus a measure that is specific to vision is important. Furthermore, the moderate, but not strong, correlations between the EYE-Q and VA and CS denote that the EYE-Q may measure other components of VI that the clinical measures overlook. These findings emphasize the importance of tools specific to the measurement of VRQOL. Therefore, the EYE-Q may be an important screening tool for eye disease and also an important measure of VRQOL. However, another consideration is that the moderate, and not strong, correlation between the EYE-Q and the clinical measures was secondary to the insensitivity of the tool's ability to comprehensively measure all aspects of VRQOL and the need for additional items with increased sensitivity to VRQOL and visual ability.

Overall, the EYE-Q demonstrated excellent internal consistency (Cronbach's α = 0.9). A reduction in items led to a 16-item questionnaire with strong internal consistency (Cronbach's α = 0.879), which may be appropriate in a general ophthalmology population. However, further validation studies in a population of children with uveitis would be necessary to determine whether there is item redundancy and a need for item reduction for this group.

Outcomes and QOL studies in JIA focus on physical disability secondary to arthritis and largely overlook impairment from eye disease (15). Visual outcome has been measured by VA and the presence of protein flare or cells during the ophthalmologic examination. However, it is important to increase our understanding of the role that VRQOL plays in children with JIA-associated uveitis in order to better assess patient-centered outcomes and the global effects of VI (psychological, physical, mental, social function). Incorporating measures of vision-specific QOL may improve measurement of outcomes, complement ophthalmologic examinations and global health measures, and augment the assessment of the impact and utility of treatment (25–27).

There is a striking paucity of validated measurement tools for VRQOL in children; therefore, it is important to develop instruments that correlate clinical measures of vision and functional ability that allow us to determine the impact of visual dysfunction and that help direct treatment and services. The adult-based NEI-VFQ is a well-validated measure of vision-related QOL but may be inappropriate for younger children, since many items are specific to adult activities (i.e., hobbies such as using hand tools or fixing things around the house, reading medicine bottles or legal forms, figuring out the accuracy of bills, or shaving and makeup use). Likewise, there are 5 published questionnaires that assess visual function in children. The Children's Visual Function Questionnaire is a parent-administered measure for young children ages ≤7 years that focuses on: 1) general health, 2) general vision, 3) competence: the ability to perform ordinary “everyday” tasks, 4) personality: effect on the child's social behavior and personality, 5) family impact: effect on the parents or the rest of the family in terms of stress and worry, and 6) treatment: effect of treatment on child and family (19). The LV-Prasad Questionnaire assesses the self-reported functional abilities of visually impaired children who are ages 8–18 years in India (18). Nineteen items assess difficulty in performing tasks in relation to distance, near, and color vision and visual field, and an item on the child's global self-assessment of vision. The Cardiff Visual Ability Questionnaire for Children measures the visual disability of children ages 5–18 years, and the Impact of Vision Impairment for Children measures the impact of VI on QOL in 8–18-year-olds (28, 29). The Vision-Related Quality of Life of Young Children and Young People is limited for use in those who are ages 10–15 years (30). These are promising instruments that are undergoing further validation but are not specific to JIA-associated uveitis. Some of the items may not apply to children in the US because of differences in the US, UK, and Australian educational structures and content. Therefore, our child-specific comprehensive EYE-Q may offer significant improvements over existing measures and also over tools for adults (14, 18, 19, 28, 29, 31).

Several limitations to this study are present. The first is that a relatively small number of children were assessed, and although 50% had normal vision, the remaining 50% had varied VI. A second limitation is the use of the Snellen chart to measure distance vision instead of the Early-Treatment Diabetic Retinopathy Study (ETDRS) acuity, which is the gold standard in ophthalmologic clinical trials (32). However, our data are analyzed using a range of best-corrected VA, which would allow for small differences in measurement, and the ETDRS acuity is not used routinely in pediatric ophthalmology clinics. Third, there was a small sample of children ages 16–18 years (11.7%). However, this is one of many steps needed to develop an instrument specific to children with JIA-associated uveitis. Prior to focusing this instrument on this population, it is important to determine whether this tool is relevant for children with and without ocular disease and with diverse VI.

Our future studies aim to administer this instrument to a larger number of children and validate its use specifically for those with uveitis. We aim to create a parent proxy module to compare to the self-report and to have available for children who are unable to complete their own instruments for whatever reasons and who are younger than 8 years old. Furthermore, we plan to assess a more comprehensive role for the EYE-Q in looking at the longitudinal course of JIA-associated uveitis.

The continued improvement of our outcome measure, the EYE-Q, will help us better determine the effects of uveitis, complications, and treatment. This is especially important in children with JIA-associated uveitis, since measures of visual function have depended primarily on the clinical examination. Development of a child-friendly, reliable, and valid instrument that assesses visual functioning within the context of QOL would enhance our understanding of the impact of complex multisystem childhood rheumatic diseases on a child's QOL and provide an opportunity for us to explore the impact of vision QOL on other complex diseases that impair vision.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Angeles-Han had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Angeles-Han, Harrison, Lehman, Hutchinson, Prahalad, Drews-Botsch.

Acquisition of data. Angeles-Han, Robb, Shainberg, Ponder, Lenhart, Hutchinson, Srivastava, Lambert.

Analysis and interpretation of data. Angeles-Han, Griffin, Harrison, Leong, Prahalad.

Ancillary