Combination Product Risk-Benefit Analysis

Abstract

Risk-benefit analysis for combination products presents challenges that neither pure drug assessment nor pure device assessment methods were designed to address. A drug-eluting stent's benefit depends on the interplay between the antiproliferative drug's elution kinetics and the metallic scaffold's mechanical properties -- the risk of late stent thrombosis is partly a drug effect and partly a device effect, and the benefit of restenosis prevention cannot be attributed to either component alone. An inhalation combination product's risk profile includes device usability failures that expose the patient to either overdose or underdose of the active drug. Theranostic combinations carry the risks of both the diagnostic agent and the therapeutic payload, linked by a targeting mechanism whose risk-benefit profile depends on the specificity of the linkage. This study presents the Combination Product Risk-Benefit Assessment Framework (CPRBAF), evaluating five risk-benefit analysis methodologies -- structured benefit-risk frameworks based on BRAT and EMA BRA templates (BRAT-EMA), multi-criteria decision analysis (MCDA), quantitative benefit-risk modelling (QBR), real-world evidence-augmented risk-benefit assessment (RWE-BRA), and AI-assisted risk-benefit synthesis (AI-RBS) -- across four combination product contexts: drug-eluting vascular stents, drug-device wound care combinations, inhalation device-drug combinations, and diagnostic-therapeutic combinations (theranostics). Performance was scored using the Risk-Benefit Effectiveness Score (RBES), a weighted composite of evidence quality (0.30), regulatory acceptance (0.20), transparency (0.20), implementation feasibility (0.15), and stakeholder communication quality (0.15). Structured benefit-risk frameworks (BRAT-EMA) achieved the highest RBES (0.899) through the best regulatory acceptance (0.940) and high transparency (0.900), supported by explicit regulatory endorsement from both FDA and EMA. MCDA ranked second (0.882) with the best transparency (0.920) and stakeholder communication (0.900). RWE-augmented assessment (0.879) achieved the best evidence quality (0.920) for post-approval risk-benefit reanalysis. AI-assisted synthesis (0.861) showed promise in literature-scale evidence integration but faces regulatory acceptance barriers