Methods: We conducted a retrospective cohort study of subjects with NMOSD from November 2005 until December 2015 using the Pediatric Health Information System. We included subjects >1 and <25 years at time of initial documentation with ICD-9-CM code for NMO who received ≤1 course of glucocorticoids. We excluded subjects with ≥1 ICD-9-CM code for multiple sclerosis. The primary exposure was ≥1 rituximab dose during initial hospitalization. The primary outcome was first re-hospitalization within 12 months. Only the first re-hospitalization that occurred >30 days later was considered (given possibility of early re-admission for repeat dosing) with data censored at 12 months. The association between rituximab and re-hospitalization was analyzed using Kaplan-Meier survival curves, log-rank test and Cox proportional hazards models. Both univariate and multivariate Cox regression models (accounting for demographics, ICU status, first-line glucocorticoid exposure and MRI studies) were calculated. Secondary outcomes included time to first re-hospitalization and duration of re-hospitalization.

Results: We included 202 subjects (73% female, 38% Caucasian) with a median age of 13 years (IQR: 10.0, 15.0). Fifty-five subjects (27%) received rituximab at first hospitalization with an increasing trend in rituximab use over the study period (p<0.01). In unadjusted models, there was no statistically significant difference in risk of re-hospitalization between those exposed and unexposed (14.6% vs. 11.6%, respectively; p=0.43). Subjects exposed to rituximab were re-hospitalized a median of 60 days (IQR: 47, 121) after initial hospitalization, while subjects not exposed to rituximab were re-hospitalized a median 125 days (IQR: 84, 187), but this was not statistically significant (p=0.06). Median duration of re-hospitalization was 2 days regardless of rituximab exposure (p=0.72). In multivariate analysis, the hazard ratio of re-hospitalization for exposed children was 1.73 (95% CI: 0.69, 4.38; p=0.25). Children exposed to glucocorticoids had significantly reduced odds of readmission (HR: 0.28, 95% CI: 0.09, 0.87; p=0.03). Additionally, subjects who had a MRI of the spine were approximately four times more likely to be re-admitted (HR: 4.29 [95% CI: 1.30, 14.13]; p=0.02).

Conclusion:  Among children with NMOSD, first-line glucocorticoid exposure but not rituximab was associated with a significantly decreased risk of re-hospitalization. However, the possibility of residual confounding by indication cannot be entirely excluded. Spinal involvement was an independent predictor of re-hospitalization.