Appraisal of computerized medical histories: comparisons between computerized and conventional records

Area of Science:

• Medical Informatics
• Internal Medicine
• Clinical Documentation

Background:

• Automated patient history systems are emerging tools in internal medicine.
• Evaluating the diagnostic accuracy and physician agreement with these systems is crucial for adoption.
• Comparison with traditional written medical records is necessary to understand their utility.

Purpose of the Study:

• To compare diagnostic statements generated from automated patient histories versus written medical records.
• To assess intra- and interobserver variability in diagnostic hypothesis generation.
• To investigate internists' opinions on the usability of different record types.

Main Methods:

• Three internists evaluated 72 records (from 18 patients) for automated histories, written records, and transcribed computerized records.
• Diagnostic hypotheses were generated for each record type.
• Intra- and interobserver agreement were calculated.
• Physician usability feedback was collected.

Main Results:

• Intraobserver agreement was highest for written records (55%), followed by automated histories (46%), and transcribed records (38%).
• Interobserver agreement was 23.5%.
• Agreement between automated histories and written records was 24%, and with transcribed records was 36%.

Conclusions:

• Automated patient histories demonstrated lower diagnostic statement agreement among internists compared to written records.
• Further development is needed to improve the consistency and reliability of automated history systems in clinical practice.
• The findings suggest current automated systems may pose challenges for consistent diagnostic interpretation.