Contributions
Abstract: P794
Type: Poster
Abstract Category: Therapy - disease modifying - 32 Others
Background: Adverse event (AE) reporting and associated treatment discontinuations are often analysed in terms of patients' characteristics. The impact of clinical site variability on AE reporting and treatment discontinuations is generally underestimated. The assessment of both patient- and site-level variability is crucial to a better understanding of AE reporting and discontinuation.
Objectives: To assess the unique contributions of patients' baseline characteristics (case mix) and clinical site variability in the occurrence of overall and GI tolerability-related AEs, and AE-related treatment discontinuation in dimethyl fumarate (DMF) clinical trials using generalized linear mixed models (GLMMs).
Methods: Pooled DEFINE and CONFIRM clinical trial data (n=2300) were used to model overall and gastro-intestinal (GI) tolerability-related AEs and related GI-AE discontinuation. Baseline predictors were age, sex, MS duration, ethnicity, smoking, alcohol consumption, BMI, prior treatment, number of prior relapses, EDSS, MSFC-4, SF-36 PCS and MCS. GLMMs were used to determine patient-level predictors and assess the variability and site-level correlation between endpoints. Residual intraclass correlation coefficients (ICC) captured the proportion of variability still attributable to sites.
Results: AEs and discontinuation were poorly predicted by patients' case mix. Site-level variability was present for both endpoints after adjusting for case mix (ICC 0.29 and 0.09). Higher GI-related AE risk was predicted by female sex, smoking, alcohol use and MSFC-4, while discontinuation from GI-related AEs was predicted by female sex and smoking. Case mix adjusted percentages (DMF vs. placebo) were 42.3 vs. 31.7 for GI-related AEs, and 1.8 vs. 0.2 (both p< 0.0001) for discontinuation due to GI-related AEs. Site-level variability was present for both these endpoints after case mix adjustment (ICC 0.12 and 0.20). Site-level residual correlations between endpoints were 0.54 for overall AEs and discontinuation, and 0.71 for GI-related AEs and discontinuation due to GI-related AEs.
Conclusions: GLMMs are a modelling tool suitable to benchmark variability and correlation between adverse events and discontinuation in MS. After adjusting for case mix, residual site level heterogeneity indicated significant variation of AEs and discontinuation by site. Site characteristics should be explored to better understand their influence on AE reporting and treatment discontinuation.
Disclosure: Supported by: Biogen, Inc.
Author disclosures:
Fabio Pellegrini, Ulrich Freudensprung, Carl de Moor and Katherine Riester are employees of, and stockholders in, Biogen.
Francesca Bovis has nothing to disclose.
Maria Pia Sormani received consulting fees from Biogen, Novartis, Genzyme, Roche, Teva, GeNeuro, Merck Serono and Medday.
Massimiliano Copetti received consulting fees from Biogen, Teva and Eisai.
Abstract: P794
Type: Poster
Abstract Category: Therapy - disease modifying - 32 Others
Background: Adverse event (AE) reporting and associated treatment discontinuations are often analysed in terms of patients' characteristics. The impact of clinical site variability on AE reporting and treatment discontinuations is generally underestimated. The assessment of both patient- and site-level variability is crucial to a better understanding of AE reporting and discontinuation.
Objectives: To assess the unique contributions of patients' baseline characteristics (case mix) and clinical site variability in the occurrence of overall and GI tolerability-related AEs, and AE-related treatment discontinuation in dimethyl fumarate (DMF) clinical trials using generalized linear mixed models (GLMMs).
Methods: Pooled DEFINE and CONFIRM clinical trial data (n=2300) were used to model overall and gastro-intestinal (GI) tolerability-related AEs and related GI-AE discontinuation. Baseline predictors were age, sex, MS duration, ethnicity, smoking, alcohol consumption, BMI, prior treatment, number of prior relapses, EDSS, MSFC-4, SF-36 PCS and MCS. GLMMs were used to determine patient-level predictors and assess the variability and site-level correlation between endpoints. Residual intraclass correlation coefficients (ICC) captured the proportion of variability still attributable to sites.
Results: AEs and discontinuation were poorly predicted by patients' case mix. Site-level variability was present for both endpoints after adjusting for case mix (ICC 0.29 and 0.09). Higher GI-related AE risk was predicted by female sex, smoking, alcohol use and MSFC-4, while discontinuation from GI-related AEs was predicted by female sex and smoking. Case mix adjusted percentages (DMF vs. placebo) were 42.3 vs. 31.7 for GI-related AEs, and 1.8 vs. 0.2 (both p< 0.0001) for discontinuation due to GI-related AEs. Site-level variability was present for both these endpoints after case mix adjustment (ICC 0.12 and 0.20). Site-level residual correlations between endpoints were 0.54 for overall AEs and discontinuation, and 0.71 for GI-related AEs and discontinuation due to GI-related AEs.
Conclusions: GLMMs are a modelling tool suitable to benchmark variability and correlation between adverse events and discontinuation in MS. After adjusting for case mix, residual site level heterogeneity indicated significant variation of AEs and discontinuation by site. Site characteristics should be explored to better understand their influence on AE reporting and treatment discontinuation.
Disclosure: Supported by: Biogen, Inc.
Author disclosures:
Fabio Pellegrini, Ulrich Freudensprung, Carl de Moor and Katherine Riester are employees of, and stockholders in, Biogen.
Francesca Bovis has nothing to disclose.
Maria Pia Sormani received consulting fees from Biogen, Novartis, Genzyme, Roche, Teva, GeNeuro, Merck Serono and Medday.
Massimiliano Copetti received consulting fees from Biogen, Teva and Eisai.