PATIENT RECRUITMENT USING ELECTRONIC HEALTH RECORDS UNDER SELECTION BIAS: A TWO-PHASE SAMPLING FRAMEWORK

Area of Science:

• Biostatistics
• Clinical Research Methodology
• Health Informatics

Background:

• Electronic health records (EHRs) are valuable for cost-effective patient recruitment in clinical research.
• Optimal cohort selection from large EHR databases for specific scientific questions remains a challenge.
• Auxiliary covariates in EHRs can improve efficiency in downstream analyses for expensive outcomes.

Purpose of the Study:

• To propose an optimal two-phase sampling design leveraging EHR auxiliary covariates for efficient cohort selection.
• To address potential selection bias inherent in EHR data for multiphase sampling.
• To improve the efficiency of clinical research analyses compared to traditional random sampling.

Main Methods:

• Developed a novel two-phase sampling design utilizing predictive auxiliary covariates from EHR data.
• Extended existing two-phase sampling literature to account for EHR selection bias.
• Validated the proposed method through simulation studies and a real-world application.

Main Results:

• The proposed optimal two-phase sampling method demonstrated significant efficiency gains over random sampling.
• The method effectively accounts for potential selection bias in EHR data.
• The application evaluating hypertension prevalence showed the practical utility of the design.

Conclusions:

• The proposed optimal two-phase sampling design enhances the efficiency of clinical research cohort selection from EHRs.
• This approach provides a robust strategy for mitigating selection bias in EHR-based research.
• Leveraging EHR covariates offers a powerful tool for improving the cost-effectiveness and statistical power of clinical studies.