Big bad data: law, public health, and biomedical databases

Area of Science:

• Public Health
• Health Informatics
• Health Policy

Background:

• Electronic health record (EHR) systems are increasingly adopted, impacting public health research and surveillance.
• While EHRs offer significant public health benefits, their data present potential limitations for researchers and analysts.
• The transition to electronic health records necessitates a critical evaluation of data quality and its implications.

Purpose of the Study:

• To examine the implications of electronic health record (EHR) adoption for public health research and surveillance.
• To highlight the potential limitations and biases within EHR data.
• To propose interventions for addressing data analysis challenges in the era of electronic big data.

Main Methods:

• Literature review and critical analysis of EHR data quality issues.
• Examination of biases including selection, confounding, and measurement bias in EHR data.
• Discussion of the impact of "big data" processing on public health findings.

Main Results:

• EHR data can be erroneous, miscoded, fragmented, and incomplete due to factors like clinician workload and poor user interface design.
• Public health findings derived from EHR data are susceptible to selection bias, confounding bias, and measurement bias.
• Automated processing of large EHR datasets without human checks exacerbates data quality concerns.

Conclusions:

• Careful consideration of EHR data limitations is essential for valid public health research and surveillance.
• Regulatory and other interventions are needed to mitigate data analysis difficulties and prevent unsound public health policies.
• Ensuring the accuracy and reliability of EHR data is critical for informing evidence-based public health strategies.