An efficient data integration scheme for synthesizing information from multiple secondary datasets for the parameter inference of the main analysis.

This article focuses on the development and evaluation of a novel statistical method called Multiple Information Borrowing (MinBo) designed to improve parameter estimation efficiency in a main analysis by integrating multiple secondary datasets containing secondary outcomes and covariates. MinBo employs empirical likelihood-based weighting schemes to borrow information from correlated secondary outcomes within the same study, offering three integration schemes—averaging, aggregating, and an omnibus approach—that adapt to varying degrees of association among secondary datasets. Simulation studies demonstrate that MinBo yields consistent and robust estimators with improved efficiency compared to traditional maximum likelihood estimation and existing single-secondary-dataset methods, even under model misspecification and partial data observation. The method is applied to the Atherosclerosis Risk in Communities (ARIC) study to identify risk factors for essential hypertension, showing that MinBo's integration of multiple secondary outcomes enhances statistical power and precision in estimating risk-factor effects.