A hospital network includes the set of hospitals linked together by patients shared. Therefore, if a patient goes to hospital “A” and then hospital “B”, these two hospitals are linked by an edge. When multiple hospital visits are included, the resulting network is an undirected network with weighted edges.

One data source to create such a network is the Healthcare Cost and Utilization Project (HCUP) databases. HCUP is derived from administrative data and contain inpatient encounter-level, clinical and nonclinical information including all-listed diagnoses and procedures, discharge status, patient demographics, and charges for all patients.

I will provide a description of the HCUP data for California, a state strategically picked due to its diverse patients who usually seek care within the state. The available data are from 2003-2011 with ~10.4 million visits contained. There is a verified patient identifier variable to track unique patients across hospital visits. This variable allows analysis of patient mobility, hospital connectivity, and changes over time.

Open research questions include: How does a hospital’s central position in the network influence inpatient mortality independent of usual factors? How does network typology differ across common acute conditions (stroke versus myocardial infarction)? How does the network change when new policies such as trauma certifications are instituted? What are the network effects on healthcare cost?