Objectives: Little is known about the real-world outcomes of bezlotoxumab, a monoclonal antibody for reducing recurrence of Clostridioides difficile Infection (CDI), in the Medicare population. This study aimed to compare CDI recurrence in Medicare beneficiaries initiating bezlotoxumab plus SoC antibiotics (bezlotoxumab group) vs. SoC antibiotics alone (control group).

Methods: A retrospective claims-based study using 2017-2018 100% Medicare Parts A, B, and D claims was conducted. The sample included elderly fee-for-service Medicare beneficiaries with a claim for bezlotoxumab plus SoC antibiotic or SoC antibiotic alone in the outpatient setting between 01/01/2018 to 09/30/2018 and evidence of clinical resolution with the initial SoC antibiotic. CDI recurrence was defined as evidence of a new SoC antibiotic fill or CDI-related hospitalization within 12 weeks of the initial SoC antibiotic completion. Propensity-score (PS) matching was the primary analytic approach.

Results: Before PS-matching, the bezlotoxumab group (n=124) was similar in age and gender but more likely to be immunocompromised (36.3% vs. 30.6%) and have higher number of CDI-related outpatient claims (mean 14.8 vs. 5.1) and SoC antibiotic prescriptions (mean 3.6 vs. 0.7) in the 12-month pre-index period than the control group (n=15,330). PS-matching resulted in 124-matched pairs that were balanced on all covariates except for small differences in region, number of comorbidities, and initial SoC antibiotic agent. Before PS-matching, CDI recurrence rates were 32.3% and 21.3% for the bezlotoxumab and the control group, respectively (p<0.05). After PS-matching, the recurrence rates for the bezlotoxumab and control group were similar at 32.3% and 33.1%, respectively (p=0.892). Sensitivity analyses with regressions controlling for the imbalanced covariates resulted in similar findings.

Conclusions: Our study did not reveal differences in CDI recurrence rates between Medicare patients receiving bezlotoxumab vs. SoC alone. However, further research is warranted given our study limitations (e.g., unobserved confounders, operationalization of CDI recurrence outcome in claims data).