Addition of constitutional symptoms to the SLEDAI-2K improves overall disease activity assessment: A pilot study

Objective: Constitutional symptoms (fatigue, lymphadenopathy, and weight loss) are not included in the SLE disease activity index-2000 (SLEDAI-2K). In this pilot study, we assessed the concurrent and construct validity of a revised SLEDAI-2K (SLED-R) that included these symptoms with the original SLEDAI-2K (SLED-O), using the physician global assessment of disease activity (PGA) as the reference.

Methods: Our revised SLED-R substituted the SLED-O's fever descriptor with a constitutional descriptor that included fever, fatigue, lymphadenopathy, and/or weight loss. SLED-O, SLED-R, PGA and patient global assessment (PtGA) scores were collected prospectively. Bland-Altman correlations for repeated measures were calculated and Meng's z-test was used to compare correlations between dependent and overlapping correlation coefficients. Associations between constitutional symptoms and disease activity measures were analyzed using Mann-Whitney U, Kruskal-Wallis, Chi-square tests and repeated measures correlations.

Results: 1123 SLED-O, SLED-R, PGA, and 1066 PtGA were collected in 239 subjects. The new descriptor was scored in 45 subjects (18.8%) and 92 instances (8.1%), while the original descriptor, fever, was scored in only 4 subjects (1.7%) and 5 instances (0.4%). Mean SLED-O, PGA and PtGA scores were higher when the constitutional descriptor was scored versus not (p < .001). The correlation between SLED-R and PGA was marginally higher than between SLED-O and PGA (p < .001). Fatigue contributed most to this increase (p = .001) and associated with both higher PGA and PtGA scores (p < .001). Mean SLED-O and PGA scores were higher when ≥1 constitutional symptom(s) were scored versus not (p < .002). Correlations between PGA and PtGA when the new descriptor was scored versus not were similar (p = .860). The frequency of concordance between PGA and PtGA was lower when the new descriptor was scored (55%) versus not (72.5%), with PGA > PtGA when the new descriptor was scored (p < .001).

Conclusion: The addition of constitutional symptoms to SLEDAI-2K, particularly fatigue, resulted in a marginal increase in its correlation with PGA, and new constitutional symptoms associated with higher SLED-O and PGA scores. As fatigue is subjective and difficult to attribute to SLE, its validity and inter-rater reliability in scoring remains uncertain. The clinical utility of SLED-R remains unclear, and further studies of its validity and reliability are needed.