Can Information Sharing Reduce Diagnostic Disparate Impact? Evidence from a Health Information Exchange

The literature has documented qualitative evidence of disparities in physician diagnoses, which can lead to disparities in healthcare delivery, patients’ perception of care, and health outcomes. In this research, we seek to understand whether (a) disparities in physician diagnoses can be attributed to disparate impact based on patient race, and (b) information technology–enabled health information sharing among healthcare providers can mitigate such disparate impact. Our empirical context focuses on racial disparities in diagnoses of heart disease between Hispanic patients and non-Hispanic, White patients. Utilizing patient-level, emergency room (ER) encounter data from 2015 to 2022, we find statistical evidence of diagnostic disparate impact where the likelihood of Hispanic patients being diagnosed with heart disease is around three percentage points lower than White patients after accounting for their underlying race-specific risk of heart disease. However, we find that health information sharing can reduce the level of diagnostic disparate impact against Hispanic patients by 18% and the likelihood of severe disparate impact by seven percentage points. We evaluate the robustness of our results using a range of specifications, such as instrumental variable estimation, falsification tests, and alternative measures of disparate impact. We also highlight the underlying mechanism that explains the role of health information sharing in mitigating diagnostic disparate impact. Specifically, we show that health information sharing between healthcare providers can reduce diagnostic uncertainty, especially for Hispanic patients, and low-skilled physicians benefit more from health information sharing compared with highly skilled physicians. This paper was accepted by D. J. Wu, information systems. Funding: The authors gratefully acknowledge the McCombs Dean’s Excellence Research Grant for this research. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.05617 .

• 저널Management Science
• 토픽scientometrics and bibliometrics research · Statistics, Probability and Uncertainty
• DOI10.1287/mnsc.2024.05617
• 저자Minghong Yuan, Indranil R. Bardhan