Racial Bias in Electronic Health Record Data: Promise, Peril, and Paths Forward

Electronic Health Records (EHRs) have revolutionized the way healthcare data are collected, stored, and analyzed. Their promise lies in the ability to capture real-time, longitudinal patient information, offering researchers and clinicians unprecedented opportunities to improve care and conduct evaluation research. Yet, beneath this promise lurks the risk that EHR data may amplify racial bias, leading to flawed research findings and inequitable health outcomes.

How Racial Bias Creeps into EHR Data

As electronic versions of existing record keeping systems and processes EHRs are not immune to the biases present in the healthcare system. Data quality issues such as missingness, misclassification, and inconsistent documentation that disproportionately affected members of racial and ethnic minority groups before the use of EHRs continue to exist. For example, the collection of race and ethnicity information varies widely across health systems, with non-Hispanic White and Black individuals showing higher rates of concordance between EHR and survey data, while other groups (such as American Indian, Alaska Native, and Asian patients) experience much lower accuracy. This means that research relying on EHR data may systematically underrepresent or misclassify certain populations, obscuring true health disparities.

Structural and Systemic Biases

Bias in EHR data arises at multiple levels. At the healthcare system level, access to care, documentation practices, and policies can skew which patients are included in the EHR and how their data is recorded. For instance, EHR systems tend to overrepresent individuals who are more likely to seek care such as women, elderly, White, and more educated patients. This selection bias means that research findings based on EHR data may not be generalizable to the broader population, especially those who face barriers to accessing care.

Implicit Bias in Clinical Practice and Documentation

Healthcare providers, like all humans, are subject to implicit biases. Studies have shown that clinicians may unconsciously favor White patients over Black, Hispanic, and other  patients of color, affecting everything from communication and treatment decisions to documentation in the EHR. For example, race and ethnicity may be recorded based on visual assessment rather than patient self-identification, leading to misclassification. Additionally, algorithms and point-of-care tools that draw on EHR data can inadvertently encode and perpetuate these biases, further influencing clinical decision-making and research outcomes.

Consequences for Research and Health Equity

When racial bias is embedded in EHR data, the consequences ripple through every stage of research and evaluation. Studies may underestimate disease burden in minority populations, fail to identify disparities, or draw incorrect conclusions about the effectiveness of interventions. This can reinforce existing inequities and hinder efforts to improve health outcomes for all. For example, maternal mortality rates for Black women are dramatically higher than for White women, a disparity that is obscured when data is misclassified or incomplete.

What Can Be Done To Realize The Promise of EHR Data?

Addressing racial bias in EHR-based research requires a multi-pronged approach:

• Standardize Race and Ethnicity Data Collection: Use patient self-identification and harmonize categories across systems.

• Improve Data Quality: Focus on completeness, disaggregation, correctness, and concordance, and conduct regular audits for bias.

• Multidisciplinary Collaboration: Involve clinicians, informaticians, statisticians, and community representatives in study design and data interpretation.

• Transparency and Validation: Clearly report data quality limitations and validate findings across diverse populations and settings.

• Ethical Oversight: Ensure that research protocols and ethics boards consider the potential for bias and its impact on health equity.

Conclusion

EHRs hold great promise for advancing health research and improving care, but only if we confront and mitigate the racial biases embedded within them. By acknowledging these challenges and working collaboratively to address them, we can harness the full potential of EHR data to promote equity and better health outcomes for all.

Sources:

• Al-Sahab, B., Leviton, A., Loddenkemper, T., Paneth, N., & Zhang, B. (2024). Biases in electronic health records data for generating real-world evidence: an overview. Journal of healthcare informatics research, 8(1), 121-139.

• Gopal, D. P., Chetty, U., O'Donnell, P., Gajria, C., & Blackadder-Weinstein, J. (2021). Implicit bias in healthcare: clinical practice, research and decision making. Future healthcare journal, 8(1), 40-48.

• Goldstein, N. D., Kahal, D., Testa, K., Gracely, E. J., & Burstyn, I. (2022). Data quality in electronic health record research: an approach for validation and quantitative bias analysis for imperfectly ascertained health outcomes via diagnostic codes. Harvard data science review, 4(2), 10-1162.

• Honeyford, K., Expert, P., Mendelsohn, E. E., Post, B., Faisal, A. A., Glampson, B., Mayer, E.K., & Costelloe, C. E. (2022). Challenges and recommendations for high quality research using electronic health records. Frontiers in digital health, 4, 940330.

• Vela, M. B., Erondu, A. I., Smith, N. A., Peek, M. E., Woodruff, J. N., & Chin, M. H. (2022). Eliminating explicit and implicit biases in health care: evidence and research needs. Annual review of public health, 43(1), 477-501.