Background Differential glucocorticoid exposure and related toxicity may exacerbate racial and ethnic disparities in lupus-related organ damage and mortality. Pediatric-onset systemic lupus erythematosus (pSLE) confers an even greater lifelong burden of cumulative medication exposure than adult-onset disease. Access to care and other social determinants of health may drive differential medication use. Therefore, we sought to determine how race and social determinants of health associate with cumulative glucocorticoid exposure over time in children with SLE. We hypothesized that minoritized race and living in more disadvantaged neighborhoods would be associated with greater average oral glucocorticoid exposure among children with pSLE in the Childhood Arthritis and Rheumatology Research Alliance (CARRA) Registry.

Methods This was a retrospective cohort study of children with pSLE enrolled in the CARRA Registry between March 2017-December 2021 with a baseline enrollment visit and ≥1 follow up Registry visits as well as a valid U.S. zip code. The primary exposures were self-identified race and/or ethnicity and national area deprivation index (ADI) linked to census tract. Time-averaged mean prednisone dose (mg/day) was used as the primary measure of cumulative glucocorticoid exposure, calculated using oral prednisone- equivalent milligram doses at each Registry visit. As secondary outcomes, we also evaluated occurrence of any prednisone restarts or dose increases between Registry visits during the study period and disease activity scores (SLEDAI-2K) over time. Associations between the primary exposures and time-averaged mean glucocorticoid dose were examined using univariate and multivariable linear regression models, adjusted for covariates (insurance status, age at enrollment, sex, major organ involvement, medication use at enrollment, disease duration at enrollment, and any secondary rheumatologic disease). We used logistic regression to analyze prednisone restart or dose increase. Linear mixed effects models were used to model disease activity over time with a subject-level random intercept.

Results A racially diverse cohort of 540 children with pSLE self-identified as 11% Asian, 27% Black, 23% Latino/a, 5% Other race, 25% White, and 9% selected more than one race and/or ethnicity. The median age at diagnosis was 14.2 years (IQR 3.2–19.0), and 46% of participants were publicly insured. Black participants were more likely to be publicly insured, live in areas with higher social deprivation (ADI), and have renal disease (table 1). In univariate analyses, Black race and living in areas in the highest ADI quartile were significantly associated with higher time-averaged mean prednisone dose (unadjustedβ=3.24, 95% CI: [0.88, 5.60] and β=4.68, 95% CI: [2.30, 7.06], respectively). ADI remained significantly associated with time-averaged mean prednisone when adjusted for race and disease duration at enrollment (adjusted β=3.21, 95% CI: [0.68, 5.75]). However, further adjustment for either insurance status or renal involvement attenuated this estimate, and neither race nor the highest quartile of ADI were significantly associated with time-averaged mean prednisone in the fully adjusted model (table 2). Black race was also associated with higher odds of prednisone dose increases or restarts in unadjusted analysis (OR=1.8, 95% CI: [1.1, 2.9], p=0.022) and approached but did not reach statistical significance at the 0.05 level in the fully adjusted model (OR= 1.7, 95% CI: [0.99, 3.1], p=0.055). Black race was independently associated with higher adjusted disease activity compared to White race (adjusted β=0.94, 95% CI: [0.11, 1.78]). Compared to privately insured children, uninsured children also had significantly higher disease activity, albeit the number of uninsured children was small (table 3).

Conclusions Insurance status and renal disease were significant predictors of greater cumulative mean prednisone dose among children with pSLE in the CARRA Registry and attenuated the associations between Black race and neighborhood-level disadvantage with prednisone dose. In contrast, Black race was independently associated with higher disease activity over time, irrespective of renal disease, neighborhood-level disadvantage, or insurance status, suggesting that there may be additional unmeasured social determinants driving this association. Furthermore, given that Black participants were much more likely to be publicly insured and live in areas of higher neighborhood disadvantage, structural confounding due to the unequal segregation of Black participants into areas of high neighborhood disadvantage must be considered. Further work to understand the complex relationships between area-level segregation and individual social determinants of health or how they mediate racial disparities is needed to identify points of intervention to improve outcomes of pSLE.