378. Cost-effectiveness of an adaptive platform trial design compared to sequential conventional clinical trials for comparative drug evaluations in bloodstream infections: a simulation study

BACKGROUND: Adaptive platform trials (APTs) have become increasingly popular in recent years in infectious diseases research. However, few studies have compared APT design against conventional randomized clinical trial (RCT) design from a cost-effectiveness standpoint. We aimed to evaluate the cost-effectiveness of APT versus conventional RCTs and quantify the trade-offs involved in choosing between these designs. METHODS: We conducted a model-based economic evaluation using a two-level, hierarchical simulation model comparing two strategies: (1) APT comparing three drugs simultaneously against a single control group, and (2) three sequential 2-arm parallel group conventional RCTs (Fig 1). Cost inputs were obtained from a recently completed conventional RCT studying bloodstream infections (BSI) and a recently launched APT for Gram-negative BSI (Table 1). 1000 Monte Carlo 2nd order iterations were performed to simulate 1000 RCTs to determine empirical Type I and II error rates across several scenario analyses. [Figure: see text] [Figure: see text] RESULTS: In the base case analysis where a less stringent interim analysis stopping rule was used, and the drugs being tested were effective, APT design was associated with lower cost ($5,368,000 vs $8,655,000), shorter duration (135 vs 242 weeks), and lower mean type II error (0.086 vs 0.213) (Table 2). However, results were highly sensitive to different scenario analyses where more stringent stopping rules were applied or if the tested drugs had no true effect. Effect sizes were less precise and on average were over-estimated with APT design (Fig 2). Type I error rates were also consistently higher with the APT strategy (mean error rates of 0.20 and 0.077 using liberal and strict χ(2)(crit) values of 3.841 and 6.635 respectively) compared to conventional design (fixed at 0.05 by design) (Fig 3). [Figure: see text] [Figure: see text] [Figure: see text] CONCLUSION: We show a proof-of-concept that simulation methods can be used to compare APT and conventional RCT designs for trial planning purposes. Neither strategy was consistently superior in terms of cost-effectiveness. Trade-offs in cost, sample size, and error rates are highly scenario dependent. Choice of trial design should depend on multiple variables, including the study question, probability of efficacy of the drug, and priorities of the investigator (Table 3). [Figure: see text] DISCLOSURES: All Authors: No reported disclosures