Developing and Validating a Model for Detecting Longitudinal Inconsistencies in the Electronic Problem List

Area of Science:

• Health Informatics
• Clinical Informatics
• Patient Safety

Background:

• Problem lists are crucial for patient care but can be distorted by clinicians in different settings.
• These distortions impact healthcare quality and patient safety.
• A standardized method is needed to assess problem list accuracy.

Purpose of the Study:

• To develop and validate a model for detecting longitudinal inconsistencies in patient problem lists.
• To provide a quantitative measure for assessing the impact of interventions on problem list quality.

Main Methods:

• A reference standard was established to create a problem-list-based model.
• Real-world problem lists were analyzed to determine an optimal categorization cutoff score.
• The model was validated against patient records to classify problem lists with or without longitudinal inconsistencies.

Main Results:

• The developed model achieved approximately 87% accuracy in categorizing problem lists.
• The model demonstrated a sensitivity of ~83% and a specificity of ~89%.
• The model effectively identified longitudinal inconsistencies in patient problem lists.

Conclusions:

• A novel model can accurately quantify inconsistencies in patient problem lists.
• This tool can be utilized in problem list studies and to measure changes resulting from policy or system modifications.
• The model supports enhanced patient safety and improved quality of care through more accurate problem list management.