Latent Profile/Class Analysis Identifying Differentiated Intervention Effects

Area of Science:

• Behavioral intervention research
• Clinical trial methodology
• Health behavior modification

Background:

• Randomized clinical trials (RCTs) are gold standard for intervention effects.
• Patient heterogeneity necessitates subgroup analysis for differential effects.
• Behavioral intervention subgroups often rely on latent constructs measured by multiple variables.

Purpose of the Study:

• Illustrate latent class analysis/latent profile analysis (LCA/LPA) for clinical trial data.
• Characterize latent subgroups for exploratory subgroup analysis.
• Identify potential differential intervention effects in behavioral research.

Main Methods:

• Selected LCA/LPA to identify patient heterogeneity based on correlated variables.
• LCA/LPA controls Type I error, assesses moderator intersections, and enhances interpretability.
• Applied LCA/LPA to a case study identifying latent classes from risk scores to test health coaching effects.

Main Results:

• Identified three distinct patient classes using clinical and perceived risk measures for coronary heart disease.
• Patients with low clinical and perceived risk showed the greatest benefit from health coaching in physical activity.
• Demonstrated differential intervention effects across identified patient profiles.

Conclusions:

• LCA/LPA provides a person-centered approach to identifying patient profiles for subgroup moderation.
• Distinct patient profiles can reveal differential intervention effects in behavioral research.
• This methodology enhances the understanding of intervention efficacy in heterogeneous populations.