Measuring guideline concordance via electronic health records: a new model for estimating concordance scores

BACKGROUND: Guideline concordance is associated with improved patient outcomes. Accurately quantifying the concordance between provided care and guideline recommendations offers valuable insights into the alignment of care with established guidelines and supports proactive approaches for improving the quality of care. Traditional models for calculating guideline concordance are effective in assessing clinical performance via cohort averages. However, these models fail at the individual patient level by not accounting for past clinical activities and their timing, which may give a distorted impression of the actual alignment between guideline recommendations and received care. OBJECTIVES: To develop a model for evaluating guideline concordance that provides accurate concordance scores at the individual patient level. METHODS: The newly developed ratio model incorporates past clinical activities and their timing (ie, past clinical trajectories), resulting in accurate, patient-centred concordance scores. We discuss its advantages and limitations and showcase its performance using clinical indicators for patients with type 2 diabetes mellitus. RESULTS: The ratio model demonstrates enhanced precision in evaluating guideline concordance at the individual level and better reflects the clinical trajectory of individual patients. While primarily designed to produce accurate individual patient scores, the model is also effective for assessing clinical performance through cohort averages. The ratio model is adaptable to diverse clinical contexts requiring regular follow-up, including chronic disease management, vaccination programmes, cancer surveillance and routine health screenings. CONCLUSIONS: The ratio model provides accurate and patient-centred guideline concordance scores. The model's enhanced precision at the individual level creates opportunities for research and clinical applications, including integration into clinical decision support systems.