Content and structure of clinical problem lists: a corpus analysis

Area of Science:

• Clinical Informatics
• Health Informatics
• Medical Informatics

Background:

• Electronic health records (EHRs) commonly use patient problem lists as a proxy for clinical summaries.
• Automated methods for generating problem lists exist, but research on optimal structuring for clinical care is limited.

Purpose of the Study:

• To analyze the structure and content of Past Medical History (PMH) sections in clinical notes.
• To identify how physicians convey patient history information beyond simple problem enumeration.
• To provide data and guidelines for improving automated clinical summary generation.

Main Methods:

• Content analysis of a large corpus of clinical notes, focusing on the Past Medical History (PMH) sections.
• Examination of semantic types of information conveyed by physicians when documenting patient history.
• Analysis of how related concepts are grouped within PMH entries.

Main Results:

• Physicians convey diverse semantic information in patient history, not solely medical problems.
• Related medical concepts are frequently grouped together within a single line of the PMH.
• Traditional problem lists, which allow only coded term enumeration, do not capture this richness.

Conclusions:

• More complex representations of patient history are valuable for clinical care.
• Findings offer insights and guidelines for developing more effective automated clinical summary generation tools.
• Improving the structure of automated summaries can enhance their utility in supporting clinical decision-making.