General linear contrasts on latent variable means: structural equation hypothesis tests for multivariate clinical...

Area of Science:

• Biomedical Research
• Statistical Modeling
• Clinical Trials

Background:

• Biomedical research traditionally favors randomized controlled trials for causal inference.
• Clinical trials increasingly include unmanipulated covariates and multiple correlated endpoints, complicating analysis.
• Modern clinical trials function as multivariate longitudinal studies, necessitating advanced statistical methods.

Purpose of the Study:

• To introduce and demonstrate the latent analysis of variance (LANOVA) method for analyzing complex clinical trial data.
• To enhance rigor and clarity in interpreting multivariate longitudinal studies.
• To adapt structural equation modeling principles for clinical research.

Main Methods:

• The latent analysis of variance (LANOVA) combines structural equation modeling and analysis of variance.
• LANOVA allows testing hypotheses on latent means, analogous to traditional ANOVA.
• Models accommodate outcomes, varying covariates, and static background covariates with set-recursive causal relations.

Main Results:

• Structural equation modeling programs can estimate LANOVA models and test hypotheses.
• The approach provides a rigorous framework for analyzing complex relationships in clinical trials.
• Hypotheses relevant to patient-reported outcomes in cancer studies were successfully expressed and tested.

Conclusions:

• LANOVA offers a powerful statistical framework for analyzing multivariate longitudinal clinical trials.
• This method adds rigor and clarity to the interpretation of complex outcome patterns.
• LANOVA facilitates the testing of standard ANOVA and ANCOVA hypotheses within advanced clinical study designs.