Under-utilisation of reproducible, child appropriate or patient reported outcome measures in childhood uveitis interventional research

Uveitis is a descriptive term used for a group of rare inflammatory eye diseases which can result in structural ocular abnormalities and visual disability. Childhood uveitis, with an estimated incidence of 5/100,000 children per annum, accounts for less than 5% of all uveitis cases [1‐3]. Almost one in five affected children lose vision in at least one eye by the time they reach adulthood [4‐6]. Uveitis differs in children and adults in aetiology, natural history, management and response to therapeutics [3]. ,Childhood uveitis is particularly challenging due to the heterogeneity of clinical presentation, difficulties in diagnosing affected children, and the delicate balance between the risk of insufficient control of inflammation with the potential negative impact of immunosuppressive chemotherapy during critical periods of ocular and general development [7]. Although commonly an isolated ocular disorder, uveitis can also occur as a manifestation of a multisystem inflammatory disease [4, 7]. In terms of prevalence, the most important associated disease is juvenile idiopathic arthritis (JIA), which is diagnosed in almost half of all childhood uveitis [7‐9]. Rather than being a diagnosis per se, JIA is itself an ‘umbrella’ term for a group of idiopathic multisystem disorders which result in chronic inflammatory arthropathy [10].

Uveitis is classified anatomically, affecting anterior, intermediate and / or posterior ocular structures, through clinical assessment by an ophthalmologist using slit lamp biomicroscopy [11]. In 2005, a multinational collaborative group developed the Standardization of Uveitis Nomenclature (SUN) [11], which included ordinal scales for disease activity quantification using this approach (Table 1). Prior to this points several different non-coterminous ordinal scales were in use internationally.¹⁰ Although vision is the key outcome of interest for any ocular disorder, the SUN scales were adopted as outcome measures for disease monitoring in clinical practice, and as surrogate endpoints for the irreversible, blinding damage caused by chronic intraocular inflammation [13]. Other potential outcomes of interest for childhood uveitis, as identified through consensus work by the Multinational Interdisciplinary Working Group for Uveitis in Childhood (MIGWUC) [12], include the incidence of sight threatening structural ocular complications, and the disease related impact on the child’s visual and global function, and on quality of life. As yet, there is no uveitis specific quality of life instrument validated for use in children, and two generic instruments (the Child Health Assessment and the Child Health Questionnaires) are erroneously described by MIGWUC as quality of life metrics rather than functional assessments (Table 1) [12].

The advent of systemic immunomodulators has improved the outlook for some affected children [13]. Despite these advances, paediatric uveitis remains a blinding disease [7]. One potential obstacle to improving outcomes for affected children is the quality and heterogeneity of disease outcome measures used in studies which assess the effectiveness of new interventions, and heterogeneity of utilised measures [14]. Interventional trials are “only as credible as their outcomes” [15], and the selection of patient-oriented outcomes is key to the assessment of the efficacy of one intervention over another. Once outcome domains are identified as meaningful and appropriate, the measure used to quantify or qualify the outcome must be able to reliably and reproducibly capture a significant change in disease or health status. Outcome measure heterogeneity is a feature of both uveitis research [14] and paediatric interventional research more broadly [16, 17] and has considerable negative impact on translational research [18, 19], by acting as a barrier to the synthesis and meta-analysis needed for the generation of an evidence base to support clinical practice.

The International Committee of Medical Journal Editors (ICMJE) has developed a system of endorsement for clinical trial databases which meet certain quality requirements. In order to receive ICMJE endorsement, a trial registry must be open access; open to all prospective registrants; managed by a not-for-profit organization; able to continuously ensure the validity of registered data; electronically searchable; and must include the minimum 20-item trial registration dataset before enrolment of the first participant (full criteria available at www.​who.​int/​ictrp/​network/​trds/​en/​index.​html). Studies must also pre-identify primary outcome measures. These registries form a repository of protocol data for completed and underway studies.

We have undertaken a systematic review of outcome measures used in interventional trials of children with, or at risk of uveitis, in order to investigate metric quality and heterogeneity.

This review of outcome measures used in interventional trials involving children with or at risk of uveitis has identified considerable heterogeneity in metric type, structure, and time at which outcome was assessed. This heterogeneity may be a reflection of the poor quality of the outcome measures used: 40% of studies used non-reproducible methodologies to assess disease activity outcomes, and the outcome measures which related to dimensions such as quality of life, structural complications, or functional sequelae of uveitis, were not validated for use in childhood.

Our review searched within registries of interventional trials, rather than within the published literature, in order to overcome the obstacle of publication bias and the outcome reporting bias, in which ‘negative’ trials, and ‘negative’ outcomes are not reported respectively [26]. Consequently, studies which have not been registered on an ICMJE recognised system have not been evaluated. As registration has, since 2005, been a pre-requisite for publication within an ICJME journal, it is unlikely that our approach has resulted in significant omissions of ongoing or completed studies. Full details of study methodology were not always available through the ICMJE-recognised registries. Although we sought further details through a supplementary search of the published literature and attempted communications with registered principal investigators for methodology details, we may be missing details which clarify the reproducibility of the metrics chosen by the reported studies.

The annual registration rate of trials involving children with uveitis increased significantly after 2005, that is in the years following the publication of the SUN guidelines. There are a number of possible explanations for this: first it may be that the registration of a number of trials in planning phase were deferred until the widely-anticipated SUN consensus had reported; second, it may be that the SUN consensus itself stimulated a number of trials by providing a much needed structure, reproducibility, and specificity to the measurement of disease activity; thirdly, this could be a chance finding. We also recognise that studies which commenced prior to 2005 and which were not registered within the ICMJE may have been omitted from our review.

Rare disease interventional research faces the challenge of adequate study recruitment to ensure sufficient sample sizes, exacerbated by the measurement and selection biases at play in the study of complex disease states. Additionally, small numbers and other methodological and logistical challenges constrain researchers’ ability to undertake randomised controlled trials. Reproducible, precise and validated outcome metrics are key to study feasibility, as are surrogate endpoints. Surrogate endpoints are a particular concern in paediatric rare disease research: the developmental trajectory of children with rare disease is often disturbed, weakening our ability to prognosticate on findings reported early in the disease course. Other obstacles to paediatric rare disease research include the appropriateness of outcome measures at different developmental stages. Almost 50% of all the studies excluded children aged under 6 years old, which may be a reflection of the difficulty in capturing outcome for young children. However, younger children are at particular risk of disease: JIA typically presents in children aged under 6 years, and this multisystem disease has a uveitis prevalence of 20–30% [7, 27, 28], with young age at presentation being a risk factor for poor outcomes in JIA associated uveitis [4, 6, 28]. Despite uveitis being a frequent extra-articular manifestation of JIA, we identified 184 interventional trials of children with JIA which did not include uveitis as an outcome measure, suggesting an importance evidence gap for research on this population. This disconnect between JIA and childhood uveitis research may be another obstacle to the translation of research into clinical care.

The absence of robust evidence on the long term clinical significance of different levels of inflammation in uveitis, and the imprecision and insensitivity of the current metrics of disease activity, complicate the use of inflammation as a surrogate endpoint for visual disability. The long duration of follow up necessary to capture uveitic visual loss, and the irreversibility of uveitic visual loss, make visual function a challenging end point for interventional childhood inflammatory disease research, and highlight the importance of developing a validated outcome able to predict final visual function (a ‘surrogate endpoint’). For children with uveitis, the majority of whom (80%) [1, 7] have chronic anterior disease, disease activity is the most commonly measured outcome dimension in interventional clinical research, and is used as a surrogate endpoint for visual disability. There is evidence that chronic disease activity results in visual disability, but the dose-response relationship is unclear, particularly for those with milder degrees of inflammation. There is also absence of clarity on the key end point for disease activity: ie whether inactive disease, or significant reduction in disease activity is the more appropriate measure of effectiveness [4, 6]. Additionally, the grading scales currently used to quantify activity are subjective, semi-quantitative, non-linear, open to intra- and interobserver variability [29], and have not been validated for use in children [7]. Objective disease activity metrics would support research on the prognostic impact of disease activity, as well as providing a precise, accurate and reproducible outcome measure. Laser flare photometry, in which the scatter of light by products within the eye is measured by a desktop instrument, has been successfully used to quantify disease activity [23, 30], and in recent trials were used to assess the efficacy of adalimumab in children with chronic anterior JIA associated uveitis [13, 31]. Laser flare meters provide an objective machine-based metric of disease activity, but have had little adoption into routine clinical practice due to expense, perceived difficulty of use, limited use beyond anterior chamber assessment, and the inability of these instruments to quantify anterior chamber cell counts. Optical coherence tomography (OCT) photography, a non-contact near infra-red high resolution imaging system, which is able to detect blood cells and proteinaceous exudate within the intraocular spaces, may have a future role in the quantification of disease activity in uveitis [32‐37].

In order to impact on clinical practice, interventional trials must evaluate outcomes which are meaningful to patients as well as practitioners. There has been little work on the importance of various outcome measures to children and families with uveitis, and a paucity of validated patient reported outcome measures (PROMs) for this group. Such work is necessary if clinical trials are to translate into improvements for affected children. There is also absence of clarity within the paediatric uveitis clinical research community on PROMs, with functional assessments (which measure the impact of disease on the child’s ability across different dimensions) conflated for quality of life metrics (which enable the child to quantify the self-perceived negative impact of their disease on their life) [12].

Multiplicity of outcome measures and use of composite outcomes can dilute the ability of individual RCT findings to support clinical care, and outcome metric heterogeneity is an obstacle to the synthesis or meta-analysis of the existing literature. Inconsistency of outcome selection also leads to clinical trials with unnecessarily large sample sizes, and to reporting biases [19, 38]. The Core Outcome Measures in Effectiveness Trials (COMET) initiative, which aims to collate and stimulate resources for the development and application of core outcome sets for clinical research, has gained interest and prominence since its establishment in 2011, with sustained growth in use of COMET resources [39]. Of the 1033 references relating to planned, ongoing and completed work on determining ‘Core Outcome Sets’ currently collated on the COMET initiative website, 18 refer to eye or vision disorders [40].

In summary, our review reports on the paucity of reproducible, age appropriate and patient reported outcome measures in childhood uveitis interventional research, and significant heterogeneity of utilised outcome measures. Although the clinical research community is working on COMET supported consensus based and patient centred approaches to outcome measures in adult uveitis [41], these advances may not translate to affected children. Clinicians and researchers interested in improving outcomes for children with uveitis must identify a patient and family centred core outcome set. This will need the active involvement of children and families in priority setting, and the outcome domains will need to consider the child’s developmental stage, and the duration of follow up at which the outcome is measured. Work will then be needed to validate these objective and / or patient reported outcome measures.