Easy-BILAG: as easy as ABC?

Abstract

The BILAG-2004 is regarded as the most comprehensive disease activity index currently used in SLE. However, its complexity can limit the efficiency and accuracy of scoring, and this has both posed a barrier to its broader adoption and potentially impacted on its utility as part of the standard clinical trial endpoints. In this issue of Rheumatology, Carter and colleagues describe development and initial accuracy and usability testing of a more streamlined version of the BILAG-2004 [1]. Dubbed the ‘Easy-BILAG’, this new version is proposed as the recommended format for scoring the BILAG-2004 index in routine clinical care. The BILAG-2004 is an organ-based scale with gradations of severity based on physician intention-to-treat. It covers nine organ systems and comprises 97 items defined in an extensive glossary. Each item is scored as new/recurrence, worse, same, improving or not present during the most recent month, relative to the month prior. These scores are integrated in an algorithm resulting in each organ being assigned a grade from A to E, where A, B and C represent major, intermediate and mild disease activity, respectively; D represents previous involvement that is currently inactive; and E represent no previous activity [2]. The proposed ‘Easy-BILAG’ maintains full fidelity to the BILAG-2004, which has been well-studied in regard to reliability and construct validity [3], but with changes in scoring format intended to enhance accuracy and feasibility. These changes, prototyped based on feedback from expert clinicians, are several, and include (i) grouping of items based on their frequency to allow better visibility and prioritization of common manifestations; (ii) the ability to derive A–E grading at the time of assessment, which eliminates the opacity of the previous computer-based algorithm; and (iii) an optional self-adjudication checklist that draws attention to common scoring errors. On face value, the changes made are logical and lead to an enhanced user experience compared with the BILAG-2004. This is reflected in operational improvements, whereby the paper-case-based validation exercises described in the study found higher accuracy of the Easy-BILAG across organ domains compared with the traditional BILAG-2004 (median accuracy 96.7% vs 87.8%), achievable in shorter completion times (average 59.5 vs 80.0 min for ten cases), as well as higher levels of inter-rater agreement and better usability ratings among clinicians. Importantly, the validation exercises were deployed using various participants, including lupus specialists and general rheumatologists, rheumatology trainees and lupus nurse specialists across both tertiary and general settings, with Easy-BILAG maintaining its better performance characteristics across this range of potential real-world users. As acknowledged by the authors, however, the Easy-BILAG is yet to be formally tested in a patient-facing context, or among clinicians for whom English is not a first language; this will be important to establish to ensure generalizability of accuracy and usability in real-world settings. While this update has clear advantages over the traditional BILAG-2004, several issues remain. Firstly, the purpose and context for implementing Easy-BILAG still requires elucidation. The authors suggest its use in routine clinical care, but even with a reduced time and administrative burden, this would need justification to support uptake in time-poor clinical environments. One such incentive may arise via treat-to-target strategies, where benefits of achieving low disease activity and/or remission states are now well-established, and require, as a prerequisite, formal measurement of disease activity. However, the best validated and most widely accepted definitions of these treat-to-target states are based at present on the SLEDAI rather than the BILAG [4, 5]. Substituting BILAG-based treat-to-target states would require appropriate validation studies. Alternatively, Easy-BILAG may find its place in a research setting. The majority of observational SLE cohorts worldwide use the SLEDAI to capture disease activity, with the major advantage over the BILAG being its speed and simplicity. This is counterbalanced, however, by numerous disadvantages, including poor sensitivity to change due to binary scoring of items, debateable weighting and omission of certain features [6]. The BILAG, in contrast, captures more comprehensively the full range of disease manifestations, and has greater responsiveness in regard to both detecting improvement and flares. Thus, as Easy-BILAG represents an administratively less burdensome form of the BILAG that still maintains its advantages, it may offer advantages over the traditional form in a clinical trial setting. Currently, the BILAG forms part of the two most common trial endpoints in SLE: the SLE Responder Index, where it is used to detect worsening of activity; and the BILAG-based combined lupus assessment, where it is the main determinant of response [6]. Although not yet specifically tested for this setting, the facility of EasyBILAG holds appeal for potentially improving the accuracy and efficiency of BILAG assessment in clinical trials. E D IT O R IA L