Characterizing the dimensions of clinical practice guideline evolution

Area of Science:

• Medical Informatics
• Knowledge Representation
• Health Services Research

Background:

• Medical knowledge is rapidly advancing, necessitating frequent updates to clinical practice guidelines (CPGs).
• Clinical decision support systems (CDSSs) rely on CPGs and must be revised to reflect changes in medical knowledge.
• Understanding how recommendations evolve between guideline versions is crucial for maintaining accurate and effective healthcare tools.

Purpose of the Study:

• To formally characterize the dimensions of recommendation evolution in successive guideline versions.
• To develop a knowledge modeling approach for comparing and identifying changes in clinical practice guidelines.
• To establish a framework for analyzing the evolution of medical recommendations over time.

Main Methods:

• Representing each atomic recommendation as a rule linking a clinical condition to action plans.
• Employing subsumption-based comparisons to formally analyze differences between guideline versions.
• Evaluating the method on French bladder cancer guidelines from 2002 and 2004 revisions.

Main Results:

• Identified seven distinct evolution patterns for recommendations: No change, Action plan refinement, New action plan, Condition refinement, Recommendation refinement, New practice, and Unmatched recommendation.
• Demonstrated a formal method for quantifying and categorizing changes in clinical practice guidelines.
• Successfully applied the method to real-world guideline revisions, highlighting its practical utility.

Conclusions:

• A formal knowledge modeling approach can effectively characterize recommendation evolution in clinical practice guidelines.
• Understanding these evolution patterns is essential for updating clinical decision support systems and ensuring guideline adherence.
• The identified patterns provide a structured way to manage and track changes in medical knowledge implementation.