Predicting Risks of Increased Morbidity among Atrial Fibrillation Patients using Consumption Classes
Background: Atrial fibrillation (AF) is the most common chronic cardiac arrhythmia. Predicting the risk of complications, or associated increases in healthcare costs, among AF patients is important for effective health care management.
Methods: A bivariate regression model including a latent morbidity index is used to predict both risk of transition to higher health costs, and mortality risk over a single year. A risk scoring algorithm for predicting transition to higher cost levels is then set out which incorporates the most significant risk factors from the regression.
Results: The regression analysis shows that in addition to age and comorbidities, baseline consumption category, ethnic group, metropolitan residence and Warfarin adherence are also significant influences on progression to increased health consumption, and relevant to assessing risk. The resulting risk scoring algorithm produces a higher AUC than the widely applied CHADS2 score.
Conclusions: The utility of a bivariate regression method with a latent morbidity index for predicting transition to worsening health status among AF patients is demonstrated. A risk scoring system based on this method outperforms an established risk score.
Full Text:Subscribers Only
- There are currently no refbacks.