This web app calculates and visualises operating characteristics of one-arm clinical trials with dichotomous (binary: response/non-response) endpoints. Such trial design often appears in a variety of studies and settings, e.g. proof-of-concept, Phase II, pediatrics, rare diseases, etc. This web app helps finding the right sample size that achieves desired operating characteristics. The technical approach has been developed in the context of a platform trial of one-arm proof-of-concept studies for neurofibromatosis as part of the EU-PEARL project. For more details, see Heimann et al. (2023).

The analysis is based on confidence distributions, which are a frequentist alternative to Bayesian posterior distributions. The method is simple to apply, does not require a prior distribution, and allows one to define similar success criteria as with a posterior distribution. For an introduction on confidence distributions refer to Xie & Singh (2013). This web app currently implements only the final analysis for a trial with no interims, although interims for efficacy or futility may be defined in such trials.

Heimann, G., Jacko, P., & Parke, T. (2023). Statistical Analysis for a Phase I Platform Trial in Neurofibromatosis. Technical Report Version 1.0. May 15, 2023. Available as Statistical Analysis for NF Master Protocols.pdf in D7.4-NF-Master-Protocol Annexes at https://eu-pearl.eu/d7-4-nf-master-protocol-for-irps/

Xie, M.-G. & Singh, K. (2013). Confidence Distribution: The Frequentist Distribution of a Parameter. International Statistical Review 81, 3-39.

The ineffective response rate p₀ is the maximum of the range of response rates that are neither of medical nor of commercial interest. It can often be derived from historical data as the spontaneous response rate observed for the disease. The desired (or clinically relevant) response rate p₁ is the minimum of the range of response rates that are of medical interest and aligned with the target product profile of the experimental intervention and should clearly exceed the spontaneous response rate observed for the disease. The conclusion rule for a trial declares efficacy of the experimental intervention if (a) the area below p₀ is smaller than α and (b) the area above p₁ is larger than 1-β (i.e., the area below p₁ is smaller than β), and declares futility if the area below p₀ is larger than γ. In all other cases, no conclusion is made.