Codifying healthcare--big data and the issue of misclassification

Area of Science:

• Medical Informatics
• Health Services Research
• Observational Studies

Background:

• Electronic medical records (EMRs) have increased large observational studies in the perioperative period.
• These studies offer real-world clinical practice data, contrasting with randomized controlled trials.
• Data sources like insurance claims and EMRs are often generated for billing or documentation, not research.

Purpose of the Study:

• To highlight the utility of large observational studies in the perioperative period.
• To address the challenges and potential inaccuracies associated with secondary data sources.
• To guide researchers in using EMR data effectively for medical decision-making.

Main Methods:

• Analysis of the characteristics and limitations of large observational studies using EMR data.
• Examination of data quality issues, particularly diagnostic code reliance.
• Discussion of strategies for mitigating errors in secondary data analysis.

Main Results:

• Observational studies provide quick, cost-effective, and representative clinical practice information.
• Reliance on billing/documentation codes can lead to misclassification and flawed conclusions.
• Potential errors in secondary data sources necessitate careful consideration by researchers.

Conclusions:

• Despite potential misclassification, large databases are valuable for effectiveness research.
• Researchers must acknowledge and address data source limitations for valid inferences.
• Careful methodology can ensure meaningful outcomes from EMR-based observational studies.