ICT Diffusion and Primary Care Methodological Contribution on Clustering Methods to Partition Medical Practices in the USA

OBJECTIVES: This project presents an analysis of IT processes on variations of prescribing patterns for patients diagnoses with diabetes type II. It follows a first study on electronic billing and analyses various stages of IT processes in clinical practices. METHODS: A sample of 610 patients is extracted from the CDC physician survey (Huttin/Wong dataset 2010). RESULTS: Two Hierarchical clustering methods: Average Linkage (AL) and Ward were used and lead to a partitioning of medical records in three clusters, showing significant differences in levels of computerized clinical information (Savage Test). CONCLUSIONS: more research is needed to include other clustering techniques and a generalization with a matrix of similarity, possibly with an optimization of multi objectives. This type of research can be used for analysis and management of propagation of IT processes inside clinical systems and control for their effects on physician prescribing behaviors.