Error adjustments for file linking methods using encrypted unique client identifier (eUCI) with application to...

Area of Science:

• Public Health
• Biostatistics
• Infectious Disease Epidemiology

Background:

• Incarceration offers a critical window for HIV testing and treatment initiation.
• Continuity of HIV care post-release is essential for patient outcomes and public health.
• Accurate linkage of patient records between correctional facilities and community care is vital for program evaluation.

Purpose of the Study:

• To develop and validate a novel Bayesian procedure for accurate record linkage between correctional systems and HIV care data.
• To improve the estimation of linkage time from incarceration to community-based HIV care.
• To enhance the reliability of data used in assessing the effectiveness of HIV care transition programs.

Main Methods:

• A Bayesian statistical procedure utilizing shared and unique variables across datasets.
• Generation of K imputed datasets to account for linkage uncertainty.
• Application of Rubin's rules for combining results from multiple imputations.
• Validation using a dedicated small dataset to refine models and priors.

Main Results:

• The proposed Bayesian method accurately identifies true links among records from correctional and HIV care systems.
• This approach improves upon existing methods by leveraging all available data.
• The procedure is flexible, computationally efficient, and robust.

Conclusions:

• The developed Bayesian procedure offers a significant advancement in accurately linking disparate health records.
• This method enhances the ability to measure and improve HIV care continuity for justice-involved populations.
• The approach has broad applicability for record linkage challenges in public health and beyond.