Methods for covariate adjustment in cost-effectiveness analysis that use cluster randomised trials

Area of Science:

• Health economics
• Biostatistics
• Clinical trial methodology

Background:

• Cost-effectiveness analysis (CEA) of cluster randomized trials (CRTs) typically assumes balanced baseline covariates.
• Systematic imbalances in covariates can occur in CRTs, potentially biasing cost-effectiveness results.
• Existing statistical methods for covariate adjustment in CRTs may not adequately address these imbalances.

Purpose of the Study:

• To present and evaluate methods for adjusting covariate imbalances in the cost-effectiveness analysis of CRTs.
• To compare the performance of different adjustment methods under various simulation scenarios.
• To identify the most reliable method for cost-effectiveness analysis in the presence of covariate imbalance.

Main Methods:

• Three methods for covariate adjustment were evaluated: seemingly unrelated regression with robust standard error, a two-stage bootstrap with seemingly unrelated regression, and multilevel models.
• These methods were applied to a CRT cost-effectiveness analysis with covariate imbalance, unequal cluster sizes, and treatment-varying prognostic relationships.
• A simulation study assessed method performance using bias, root mean squared error, and confidence interval coverage of incremental net benefit.

Main Results:

• Cost-effectiveness results varied depending on the covariate adjustment method used.
• Unadjusted methods produced biased results, even with low confounding.
• All adjusted methods yielded unbiased results; multilevel models demonstrated superior confidence interval coverage, particularly with few, unequal-sized clusters.

Conclusions:

• Covariate adjustment is essential for unbiased cost-effectiveness analysis in CRTs with baseline imbalances.
• Multilevel models offer a robust and reliable approach for covariate adjustment in CRTs, outperforming other methods, especially in challenging scenarios.
• The choice of adjustment method significantly impacts cost-effectiveness findings, highlighting the importance of appropriate statistical techniques.