Abstract for: Variation in a Medical Home: Clusters of Practice, Disparate Outcomes and Dynamic Tensions
Medical home evaluations often report modest and/or mixed results. We explore the variation in implementation, impact and understanding in a medical home. We also explore the use of analytical methods based in statistics that complement dynamic analysis. This project uses a mixed methods, multi-stage design to explore medical home variation. First, we explored implementation of the model, using semi-structured interviews, ANOVA and cluster analysis. Second, we explored the relationship between implementation and impact on patient outcomes using the New York Algorithm for Emergency Department (ED) use, propensity score matching and t-tests. Third, we explored employees’ understanding of the model using causal loop diagrams (CLDs). Primary care providers practice differently across clinics and teams, and even across teams in the same clinic. This is shown to impact patient ED use, which varied significantly across clusters (at alpha = 0.05). CLDs point to key structural mechanisms -- understanding them and patience are required to overcome short term balancing loops. Dynamic tensions were found to produce variation in medical home implementation and resulting variation in outcomes. Causal Loop Diagrams identified causal pathways behind the tensions felt and outcomes observed. We are now developing the SD quantitative model.