ABSTRACT
INTRODUCTION: Comorbidity burden may be associated with treatment-effect heterogeneity (HTE) in clinical trials, which could alter the interpretation or clinical translation of results for many patients in the real world. OBJECTIVE: In this analysis, we sought to determine the distribution of multimorbidity scores in patients enrolled in SCD-HeFT (Sudden Cardiac Death in Heart Failure Trial) and tested the association between comorbidity burden and treatment efficacy for the outcome of all-cause death. METHODS: Each patient was assigned a modified Charlson Comorbidity Index (mCCI) score from 1 to 14 based on available enrollment data. We investigated the relationship between mCCI score and time to all-cause death using Cox proportional hazards models. Models were fit for quartiles of the comorbidity index, reference coding was used, with quartile 1 (Q1; mCCI score of 1-2) selected as the reference. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were reported from these models. Following the same analysis framework as the original manuscript, patients assigned to amiodarone or implantable cardioverter-defibrillator (ICD) were compared with those assigned to placebo in separate Cox models. Each model included the mCCI score in quartiles, group assignment, and an interaction term for the quartile and group assignment. HRs and corresponding 97.5% CIs were reported from these models. RESULTS: The majority of patients had an mCCI score ≤5 (75.4%), and mortality risk was associated with increasing score. The HRs for Q2 (score 3), Q3 (scores 4-5), and Q4 (scores ≥6) were 1.46 (97.5% CI 1.06-1.99), 3.03 (97.5% CI 2.35-3.90), and 4.51 (97.5% CI 3.46-5.88), respectively. For the subgroup analysis, amiodarone was not associated with a significant difference compared with placebo for individuals in Q1-Q3; however, it was associated with an increase in death for those in Q4 (HR 1.50; 97.5% CI 1.03-2.18). ICD was associated with a significant reduction in death compared with placebo for individuals in Q1 and Q3 (HR 0.42; 97.5% CI 0.20-0.84 and HR 0.70; 97.5% CI 0.50-0.97, respectively) but not for those in Q2 or Q4. Interaction testing across subgroups suggested HTE for amiodarone (p = 0.07) and ICD (p = 0.08) versus placebo across mCCI quartiles. CONCLUSIONS: Increasing comorbidity burden was associated with HTE when evaluating amiodarone and ICD compared with placebo in the SCD-HeFT trial. Our results highlight the importance of enrolling diverse patient populations in clinical trials and considering the possibility of HTE when translating results to clinical practice.
