Trial Analysis and Interpretation in Critical Care Using the Evidential (Likelihood) Approach: Rationale and...

Area of Science:

• Clinical Research Methodology
• Biostatistics
• Evidence Synthesis

Background:

• Traditional P value-based analyses in critical care often yield inconclusive results.
• Frequentist statistics are predominant, with a recent resurgence in Bayesian methods.
• The evidential (likelihood) framework offers a complementary perspective bridging inferential paradigms.

Purpose of the Study:

• Introduce the evidential framework as an analytical tool for critical care research.
• Demonstrate the application of the evidential approach using real clinical trial data.
• Facilitate the implementation of the evidential framework in scientific practice.

Main Methods:

• Applied the evidential framework to data from mechanical ventilation trials (ART, STAMINA) and a balanced solutions trial (BaSICS).
• Focused on translating concepts and terminology across statistical paradigms.
• Examined the framework's measures of effect, sequential analysis, and reporting.

Main Results:

• The evidential framework provides clinically intuitive interpretation by assessing relative evidence between hypotheses.
• Measures of effect include likelihood ratios, support values, and support intervals.
• An interactive Shiny application was developed to aid implementation.

Conclusions:

• The evidential framework offers complementary insights that align with clinical reasoning.
• It provides a valuable alternative or supplement to traditional statistical methods in critical care.
• The developed tool aims to increase the adoption and utility of the evidential approach.