On a model-based approach to estimating efficacy in clinical trials

Area of Science:

• Clinical Trials
• Biostatistics
• Pharmacoeconomics

Background:

• Treatment efficacy is paramount in Phase III clinical trials.
• Patient non-compliance with treatment regimens can significantly confound efficacy assessments.
• Existing models often assume comparability between drug and placebo compliance.

Purpose of the Study:

• To investigate the impact of non-comparable drug and placebo compliance on treatment effect estimation.
• To assess the robustness of a model-based approach (Efron and Feldman) under compliance assumption violations.
• To determine the extent of bias introduced by varying degrees of compliance non-comparability.

Main Methods:

• Utilized a model-based statistical approach incorporating drug and placebo compliance data.
• Conducted a simulation study to evaluate estimation robustness under violated compliance comparability.
• Analyzed the influence of compliance-response relationship strength on bias.

Main Results:

• Violation of the drug-placebo compliance comparability assumption can lead to severely biased treatment effect estimates.
• Moderate non-comparability (e.g., normalized compliance correlations of 0.4) demonstrated significant bias.
• The degree of bias is modulated by the strength of the association between patient compliance and placebo response.

Conclusions:

• The assumption of comparable drug and placebo compliance is critical for accurate treatment effect estimation in clinical trials.
• Researchers must carefully consider and assess compliance comparability to avoid biased findings.
• The relationship between compliance and placebo response is a key factor influencing the impact of non-comparability.