To develop a set of core outcome measures for use in randomized controlled trials (RCTs) and longitudinal observational studies in juvenile idiopathic arthritis (JIA)–associated uveitis.
To develop a set of core outcome measures for use in randomized controlled trials (RCTs) and longitudinal observational studies in juvenile idiopathic arthritis (JIA)–associated uveitis.
The literature relating to outcome measures used in studies of uveitis in childhood and adolescence was reviewed. A set of core outcomes and domains was established using the Delphi process. This was reviewed by a representative multinational interdisciplinary working group. Nominal group technique consensus was reached on face and content validity of the range and content of the domains. The outcomes and the appropriate instruments for uveitis trials were adapted to the age ranges of patients with JIA-associated uveitis.
Consensus was reached that data should be reported at defined time points in longitudinal studies with patients stratified by prognostic markers. Visual acuity testing should be age appropriate. The severity of uveitis (measured as anterior chamber cell grade) and duration of active inflammation should be documented. Visually significant structural complications should be recorded and quantified with standard measures. The responses to treatment and corticosteroid-sparing effects of treatment should be documented. Patient-reported disease activity and age-specific uveitis-related quality of life should be reported using appropriate questionnaires.
The proposed outcome measures in JIA-associated uveitis should aid in the standardization and comparison of future RCTs of the treatment regimens for this disease. The proposed outcome measures will be verified in a prospective validation study.
Juvenile idiopathic arthritis (JIA) is the most common extraocular disease associated with pediatric uveitis (1). The prevalence of JIA is 50–200 per 100,000 children (2, 3), and in 10–18% of patients, arthritis is accompanied by uveitis (3–5). JIA-associated uveitis has a wide spectrum of manifestations and severities (6–11). Risk factors for uveitis include antinuclear antibody (ANA) positivity, oligoarthritis, and early onset of arthritis. Despite advances in the treatment of childhood arthritis, JIA-associated uveitis is still a common cause of eye-related morbidity.
Observations published worldwide about the outcomes of JIA-associated uveitis are surprisingly diverse. The broad variety of outcome measures utilized makes the comparison of the disease course, risk for structural complications, levels of impairment in visual function, and responses to treatment difficult.
The treatment of JIA-associated uveitis is guided by the level of anterior chamber (AC) inflammation, risk for significant structural complications, and activity of the arthritis (12). The variety of risks for vision loss and variable course complicate the interpretation of treatment studies of JIA-associated uveitis, since significant events may occur years after any therapy.
The currently published treatment studies of JIA-associated uveitis are limited by small numbers, heterogeneous populations, and their retrospective nature. Only 1 randomized controlled trial (RCT) has been conducted in patients with JIA-associated uveitis using etanercept, which did not demonstrate efficacy (13).
The Standardization of Uveitis Nomenclature (SUN) Working Group has provided a standardized nomenclature of uveitis, inflammation grading, and outcome measures (14). This instrument is of importance for the comparison of outcomes and trials, and for the optimal design of future trials.
There are limitations to the utility of generic instruments for all categories of uveitis, since uveitis syndromes may have relatively unique patterns of inflammation. The disease course and complications of JIA-associated uveitis are relatively unique; arthritis and uveitis onset in early childhood results in problems that create an impact that may not be adequately covered with adult measures of visual function and/or health-related quality of life.
Since specific outcome measures have not been established for JIA-associated uveitis, a group of ophthalmologists and pediatric rheumatologists gathered to consider this issue. The recommendations of the SUN Working Group were reviewed, with special consideration given to their applicability for reporting clinical outcomes in studies of JIA-associated uveitis.
We proposed a core set of outcome measures for juvenile idiopathic arthritis (JIA)–associated uveitis.
The proposed outcome measures in JIA-associated uveitis should aid in standardization and comparison of future randomized controlled trials of treatments for this disease.
An informal international interdisciplinary working group for uveitis in childhood, consisting of ophthalmologists and pediatric rheumatologists, was formed. The participants had relevant experience in leading long-term prognostic studies, interventional trials, and the methodology of outcome consensus and validation techniques, as well as experience in the management of JIA and JIA-associated uveitis. The participants were from 8 European and 4 North American centers. No patients or parents, or representatives from regulatory agencies or pharmaceutical companies, participated.
The stated aims were to evaluate the currently used outcome measures in published studies, to achieve face and content validity for the relevant domains and outcome measures suitable for interventional studies, and to establish the methodology for the prospective testing of discriminant and predictive validity (15) for the proposed outcome measures.
Members of the group (AH, IF, RKS, and SN) reviewed the published literature concerning outcome measures in uveitis in childhood and adolescence. As a result of this review, a list of candidate domains and items was generated. A questionnaire with the list of domains and items was circulated to all of the members. Each domain and item was ranked on a scale from 1–5, where 1 = highest importance and 5 = lowest importance. A Delphi procedure (16) was then used to rank the domain and item scores that were returned from all of the members. A consensus meeting took place from May 22–23, 2009 in Barcelona, where 14 of the 16 members participated.
During the meeting, the ranking list of all domains and items (with mean scores and SDs) was presented to the participants (data not shown). A nominal group technique was used to develop a consensus. Briefly, during the discussion rounds, each member first individually explained their ratings of each of the domains and items. The ratings were then discussed by the entire group, and the list of domains and items and their description and intent were thereby modified to achieve consensus from the entire group. The domains or items with low importance or redundant variables were eliminated. Consensus was obtained if all of the members agreed. In the case of disagreement, even by a minority, it was considered that consensus had not been reached.
Subsequently, all of the domains and items were ranked for a second time by all of the participants. Following the analysis, the results were presented and discussed again by electronic communication. The domains and items were modified to reach consensus on outcome measures for as many domains and items as possible. The current proposal for domains and associated items was generated (Table 1). The domains and items for which no consensus was achieved were also documented.
|Grade of cells in anterior chamber||Slit-lamp examination|
|Grade of flare in anterior chamber*||Slit-lamp examination for routine clinical practice and prospective trials|
|Laser flare photometry for prospective trials|
|Number of visits with active uveitis||Records of treating physician|
|Duration of activity over a minimum of 4 visits/year|
|Visual acuity (appropriate test for age)||Best-corrected visual acuity|
|Thresholds: ≤20/50, ≤20/200, and no light perception|
|Estimate contribution of amblyopia, yes/no|
|Development of structural complications||Synechiae, yes/no|
|Initial and additional|
|Ocular hypotony, yes/no|
|Ocular hypertension, yes/no|
|Band keratopathy in the central cornea, yes/no|
|Macular edema, yes/no|
|Funduscopy for routine clinical practice|
|Funduscopy and optical coherence tomography for prospective trials|
|Epiretinal membrane formation, yes/no|
|Funduscopy for routine clinical practice|
|Funduscopy and optical coherence tomography for prospective trials|
|Quality of life||Childhood Health Assessment Questionnaire|
|Child Health Questionnaire|
|Pediatric Quality of Life Inventory|
|Uveitis-specific quality of life instrument|
|Overall uveitis-related disability||Assessment by parents, visual analog scale|
|Assessment by children, visual analog scale|
|Assessment by treating ophthalmologist, visual analog scale|
|Assessment by treating pediatric rheumatologist, visual analog scale|
|Social outcome||School/kindergarten absence|
|Antiinflammatory medication*||Reduction of corticosteroid dose|
|Biomarkers||Research tools (not currently available)|
Grading of the cell content in the AC was considered an important measure to assess uveitis activity (14). It is presumed that the effect of the antiinflammatory treatment is directed by the change in the number of cells. The SUN Working Group suggested a grading system for the estimated AC cell count (14).
Although the enumeration of the exact number of cells circulating in the slit beam may be preferred, the conference participants decided that estimation by the scale proposed by the SUN Working Group is the most appropriate for routine clinical practice. However, it is a subjective measure, and the recording of the exact cell grade may only be fair to moderate (17) and requires clinical experience, especially in young children with synechiae and media opacities.
There are also problems in determining the optimal scale for grading AC cells in JIA-associated uveitis. There are profound prognostic differences between the absence of cells and the presence of even a small number; most clinical activity in JIA-associated uveitis occurs within 3 grades of activity (0.5–2+), and so aggregated counts may lose sensitivity (18). Presently, there is no evidence of a different risk attached to these 3 levels of activity (19), and some studies have shown no prognostic significance for cell counts in the presence of flare (18).
A consensus was reached that the cell grade should be reported, and activity or inactivity be stated, at each visit. As determined by the SUN Working Group, improvement following treatment is defined as a decrease in the AC cell grade of at least 2 steps or to inactive, and worsening is defined as an increase in the AC cell grade by at least 2 steps or to the maximum grade.
The participants suggested that it is important to consider ocular complications, such as synechiae formation, macular edema, optic disc edema, and persistent ocular hypotony when interpreting AC cell grade, since the coexistence of other complications may greatly alter the prognostic significance of the levels of AC activity (19).
AC flare has been used as a measure of both the uveitis activity and tissue damage (14, 18). Higher values are commonly detected during uveitis exacerbation and often correlate with the presence of AC cells (18); however, flare may be present without cells in the AC (18) and may persist despite the disease being in remission. Only a treatment trial can differentiate reversible and irreversible flare, and again, the prognostic significance of flare may be dependent on both coexisting structural damage and the effect of the treatment administered at the time of measurement. Despite these uncertainties, the level of AC flare may be a good indicator of prognosis and a marker for damage of intraocular structures (18).
The participants believed that flare in children can be estimated only roughly (e.g., as no flare, moderate flare, and intense flare) and that fibrin formation should be documented separately. As was also recommended for AC cells, the presence of AC flare should be related to other signs of activity and tissue damage.
For the detection of AC flare, the slit lamp is widely used in clinical practice; however, flare estimates may differ between observers and visits (18, 20). Laser flare photometry is an objective technique (17, 20), but it is time consuming, expensive, and not feasible as a measure in many patients with JIA-associated uveitis. The participants therefore considered that photometry should not be a necessary instrument, but remain a research tool for prospective RCTs.
The group recommended that for longitudinal observational studies and RCTs, the results should be reported at baseline or study entry, and thereafter at well-defined time points. The duration of uveitis activity should be reported because it is likely a critical determinant for the long-term prognosis. Since uveitis activity cannot be measured between visits, at least 4 visits per year (ideally at 3-month intervals) should occur, and activity or inactivity should be reported at each visit.
Visual acuity is an important outcome measure for childhood uveitis in both epidemiologic studies and RCTs. Binocular visual acuity is related to visual function and quality of life. It was readily agreed on by the group that visual acuity of each eye should be included as a measure of severity of disease.
Testing of best-corrected visual acuity should be performed unilaterally and bilaterally. For pediatric trials, the selection of an age-appropriate test is critical. For example, Lea and E tests and HOTV distance visual acuity tests are acceptable for a child age <7 years (21), while reading numbers at a distance of 5 meters or the EDTRS chart may be more appropriate for adolescents. Visual acuity testing relies on patient motivation, which may be unpredictable in children. Visual acuity test results should be converted to logMAR units for statistical analysis.
Documentation of visual acuity is preferentially reported as visual acuity thresholds, for example, ≤20/50, ≤20/200 (which defines legal blindness in North America and most of Europe), and no light perception. The meeting participants considered that visual acuity is not a useful measurement of uveitis activity; however, since it is influenced by prevalent complications, it is a very important indicator of ocular damage related to disease.
Amblyopia is a reduction of visual acuity caused by impaired function of the central nervous system, and not by structural abnormalities of the eye. Among the causes of visual deterioration in JIA-associated uveitis are macular edema, cataract formation, unilateral cycloplegia, and noncompliance with refractive correction.
The frequency and functional consequences of amblyopia in children with uveitis are not defined well, and amblyopia is not a useful measurement of uveitis activity or severity because its incidence is age dependent. Its presence may confound the interpretation of immunosuppressive trial outcomes; however, it is a significant contributor to visual function outcomes and treatment costs. The participants reached consensus that clinicians should estimate the contribution of amblyopia to reduced vision and identify its underlying cause, where possible.
The meeting participants agreed that structural damage is an important outcome measure that reflects both previous and current disease activity. While the development of new complications may reflect disease activity, they may also occur in the absence of active uveitis (e.g., epiretinal membrane formation, band keratopathy, ocular hypotony and hypertension, and glaucoma).
Synechiae formation is a frequent complication of JIA-associated uveitis (1, 5) and is a useful indicator of disease activity and severity of inflammation, as well as a prognostic indicator of subsequent vision loss and cataract surgery (19, 22). There was consensus by the group that the development of synechiae is an important outcome measure in childhood uveitis; therefore, formation of the first or additional synechiae should be reported. The sensitivity of detection may be higher with slit-lamp photography than the description of the involved quadrants, and therefore photography may be appropriate for prospective studies. However, where extensive posterior synechiae exist, even photography has poor sensitivity for detection of new synechiae.
Cataract formation is a frequent complication of JIA-associated uveitis resulting in vision loss (1, 5, 19, 22). The participants agreed that cataract formation should be reported as an outcome measure and may be semiquantified by slit-lamp evaluation using published grading systems, with consideration of nuclear, cortical, and posterior subcapsular opacification (23).
Clinical macular edema causes vision loss in 3–26% of patients with JIA-associated uveitis (19, 24, 25). It has probably been underestimated prior to the widespread use of fundus lenses and optical coherence tomography (OCT). In 1 study, macular edema was found in 84% of cases of JIA-associated uveitis (26). There was consensus among the participants that the presence or absence of macular edema should be documented as an outcome measure. The participants agreed that macular edema is often associated with active inflammation in both the AC and vitreous, and might require interpretation in conjunction with the levels of both anterior and vitreous cells and flare.
OCT has recently been introduced as a more sensitive measure for detecting macular edema and is available in most clinics. It is noninvasive and can be performed in young children. This modality provides a reliable measurement of foveal thickness and is a sensitive and specific measure to detect improvement or reversal of macular edema during therapeutic intervention.
The panel reached consensus that for the detection of macular edema, using OCT is favored in children over fluorescein angiography, which is more invasive, may be complicated by nausea and allergic reactions, and is hampered by the lack of compliance in children. Macular edema may be present in the majority of uveitis patients within the first 3 months after intraocular surgery, and therefore any antecedent surgery should be reported when macular edema is measured.
Epiretinal membrane formation has occasionally been described in JIA patients and may be relevant to visual outcome (19). The participants agreed that the presence or absence of epiretinal membrane on funduscopy should be reported as an outcome measure. Due to the higher sensitivity for detection of membranes, OCT should be added as an ancillary measure in future prospective studies.
Band keratopathy has been diagnosed in 15–20% of JIA patients with uveitis (1, 7). The participants agreed that band keratopathy should be reported as an outcome measure only if it affects the central cornea with consequent vision loss.
Ocular hypertension and secondary glaucoma, as defined by the presence of pathologic cupping of the optic disc and/or visual field defects in the presence of intraocular pressure (IOP) ≥21 mm Hg, are common complications of JIA-associated uveitis (10–40%) and significant risk factors for vision loss (1, 7, 27).
The participants agreed that elevated IOP and a glaucomatous optic disc should be reported as separate outcome measures. In accordance with the previous publication by the SUN Working Group (14), a substantial rise in IOP of ≥10 mm Hg compared to the baseline, IOP of ≥21 mm Hg, and/or IOP of ≥30 mm Hg should also be reported. The participants agreed that glaucomatous visual field defects should be reported, since such defects may impact visual functioning; however, visual field testing in children may be influenced substantially by compliance and coexistent structural complications. Administration of IOP-lowering medications, both topical and systemic, may be documented, but are heavily dependent on the therapeutic practice of the treating ophthalmologist.
Hypotony has been found in 3–9% of patients with JIA-associated uveitis, and is associated with poor visual outcome (5, 7, 19). It was agreed by the participants that persistent hypotony with IOP ≤6 mm Hg should be reported as an outcome measure. Hypotony in childhood uveitis often correlates with active inflammation, and IOP may revert to normal when resolution is achieved after appropriate therapy.
Although vitreous haze and cells occur in JIA-associated uveitis and have been used as outcome measures in adult uveitis, they only occur in a subgroup of JIA-associated uveitis patients. There was insufficient evidence at the present time of their incidence and relationship to other signs, especially posterior segment edema, to establish their utility as an additional outcome measure.
The participants suggested that in addition to the physician's perspective on disease activity and severity, the patient's perspective, as measured with quality of life assessment questionnaires, should be reported. It is well recognized that physicians' and patients' perspectives about treatment costs and responses may differ.
Ophthalmologists should record uveitis activity, severity, and overall uveitis-related disability using a visual analog scale. Pediatric rheumatologists should provide a rating based on the overall JIA activity and severity. Uveitis-related disability should be reported by the patients or caregivers. Several health-related quality of life questionnaires that are applicable to children, but not to pediatric eye disease specifically, have been validated in languages other than English, such as the Child Health Questionnaire, the Childhood Health Assessment Questionnaire (28), and the Pediatric Quality of Life Inventory (29).
Uveitis-related quality of life assessment questionnaires, which consider complex parameters of most relevance to pediatric uveitis such as medical (topical and systemic) and surgical treatment, as well as the need for frequent monitoring by ophthalmologists and pediatric rheumatologists, were not available at the time of the consensus meeting.
Within the SUN Working Group, there was consensus that, for reporting in adult patients, reduction in dosages of prednisone to 10 mg/day while maintaining inactive uveitis be considered the primary outcome for successful corticosteroid sparing (14). No consensus was found in our group regarding what the reduction of topical and systemic corticosteroids should be defined as for a desirable outcome measure. Some participants were concerned that this was confounded by indication bias; however, the group agreed that it is important information to be collected in trials. In addition to the previous cumulative dose, the doses at baseline and reductions to below ≤0.1 mg/kg in the course of a study should be reported. No consensus was reached regarding the clinically relevant level of reduction in dose of other immunosuppressants that should be recorded.
Patients with JIA-associated uveitis frequently require cataract surgery, vitrectomy, and glaucoma surgery (9, 19, 22, 30). The participants agreed that any surgery at baseline and/or after study entry should be documented as an indicator of severity of disease and tissue damage. Surgery is not a primary outcome measure because indications for surgery and the surgical methods used are biased by the practice patterns of the treating ophthalmologist. In addition, the need for surgery often depends on the prior course and severity of the disease, and therefore it is not a useful measure of immunosuppressive effectiveness. Surgery is generally performed after inflammation has been controlled and it may induce further inflammation; therefore, it may confound activity outcome studies, but it has great significance for both patient treatment costs and health utilization outcomes.
There was consensus among the participants that sensitive and reliable biomarkers have not been adequately studied. Future opportunities may include gene expression and proteomic profiling of the serum, peripheral blood leukocytes and aqueous humor, measurement of acute-phase reactants, HLA typing, and determination of ANAs.
The curiosity about outcomes has to be balanced with the burden of research for patients and the practicality of acquiring complete data sets. The primary purpose of this meeting was to determine a minimum outcome set, enabling multisite comparisons, rather than providing a comprehensive data set for research. The majority of the proposed items are part of the routine monitoring of patients with JIA-associated uveitis and the additional costs of providing global assessments and health-related quality of life measures are feasible in JIA and have become routine in many centers.
The aim of a core set should be to provide the minimum data required to establish significant differences and demonstrate congruence between patient, researcher, and regulator perception of those end points. It is highly likely that the multiple structural and activity changes recorded in JIA-associated uveitis can be simplified for the purpose of evaluating the efficacy of treatment. Without prospective studies of how these change over time, it is impossible to evaluate redundancy and efficacy from the contemporary literature.
The aim of this meeting was to develop a preliminary core set of outcome measures for JIA-associated uveitis. Trials are needed to determine the benefits and costs of health interventions and require tools to measure changes in relevant outcomes. Standardizing outcomes allows data pooling, and a core set of outcomes, agreed upon by both researchers and patients, provides a common focus for interventional studies. Disease-specific and universally agreed upon outcomes are likely to reduce selective reporting and reporting bias (31). The European Medicines Agency now requires pediatric trial end points to be appropriate and contained within the pediatric investigation plans.
The recommended process of defining outcomes is to form a group of relevant stakeholders, assess available evidence, and agree upon selected outcomes through a structured consensus-generating process. The optimal process is not established (32). The Delphi process of sequential questionnaires avoids the possible biases introduced by domineering characters that may participate in face-to-face discussion, but can be biased by the initial choice of participants and areas of interest. In the absence of an international body specifically entrusted with guideline development for JIA-associated uveitis, this study is open to the criticism that the choice of the participants was necessarily arbitrary. In the future, patients and caregivers, and industry participants, will also be incorporated in the process.
The content of the preliminary outcome sets should be driven by the questions prioritized by stakeholders rather than being constrained by the existing literature, especially where no outcomes have been defined and a literature meta-analysis is not feasible.
Pediatric core outcomes are particularly scarce (33) and have unique problems (e.g., requiring data acquisition from parents and disease impact on caregivers as well as patients). Examination of visual function appropriately age adjusted for relevant activities and reporting skills (34, 35) can help expand the utility of quality of life measures. Multiple domains are necessary to capture all aspects of such a disease. Most recently, a vision-related quality of life questionnaire was developed for children with JIA-associated uveitis (35).
Another shortcoming of this report is the lack of validation beyond face and concept validity. However, the primary purpose of the meeting was to choose outcomes and suggest the feasibility of measurement rather than analyze the responsiveness, reproducibility, and redundancy of these instruments, all of which require further data collection. Retrospective analysis of redundancy was used for the measures available in a single cohort of patients. Prospective validation of patients undergoing treatment changes has been used previously to test proposed outcomes in pediatric systemic lupus erythematosus (36) and was the method chosen to pursue the validation process.
For a majority of the domains and items, consensus was obtained. This result now must be tested, according to the Outcome Measures in Rheumatology requirements, for truth, discrimination, and feasibility (15). A meeting to revise the current proposal is planned, following the analysis of data from ∼100 children with JIA-associated uveitis who have completed at least 3 consecutive visits after initiation of disease-modifying antirheumatic drugs.
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. Heiligenhaus 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. Heiligenhaus, Foeldvari, Edelsten, Saurenmann, Bodaghi, Graham, Anton, Kotaniemi, Minden, Nielsen, Ramanan, Strand.
Acquisition of data. Heiligenhaus, Foeldvari, Smith, Saurenmann, de Boer, Graham, Anton, Kotaniemi, Mackensen, Nielsen, Rabinovich, Ramanan.
Analysis and interpretation of data. Heiligenhaus, Foeldvari, Bodaghi, Mackensen, Rabinovich.
Abbott Laboratories Immunology Division had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Abbott Laboratories Immunology Division.