Online Program

Geospatial Clustering of Hospital Readmission Rates in the North Carolina Triangle Area

Tuesday, November 3, 2015 : 8:50 a.m. - 9:10 a.m.

Michaela Dinan, Ph.D, Duke Clinical Research Institute, Duke University, NC, NC
Ben Strauss, MS, Children's Environmental Health Initiative, Durham, NC
Kevin Schulman, MD, Duke Clinical Research Institute, Duke University, Durham, NC
Marie Lynn Miranda, PhD, School of Natural Resources & Environment, University of Michigan, Ann Arbor, MI

Hospital readmissions are a major source of preventable U.S. health care expenditures that have been previously associated with patient demographics, clinical factors, neighborhood characteristics, and geographic variation in health care systems. 


We examined whether patients at elevated risk of readmission are spatially clustered after adjusting for patient-level demographic and clinical factors within a region largely served by a single academic health system in North Carolina.


We conducted a retrospective geospatial analysis of admissions (N=102,807) between 2008 and 2011 from patients at one of three major Duke University Health System hospitals.  Readmission was defined as a second admission within 30 days of discharge.  Patterns in readmission were investigated using a sequential application of aspatial, spatial, and a novel combination of aspatial and spatial factors using an intrinsic conditional autoregressive (ICAR) model.


Overall, 12,124 (11.8%) of index hospitalizations were followed by a readmission within 30 days of discharge. Non-spatial multivariable analysis revealed increased odds of readmission strongly associated with longer length of stay and older age, followed by black race and primary diagnosis. s.  We observed persistent spatial clustering of increased readmission rates within low socioeconomic status regions in central Durham after adjusting for patient and admission-level variables including length of stay, race, primary diagnosis, and age. Decreased rates of readmission were clustered within central Wake County. Combined spatial clustering analysis showed persistent spatial clustering of increased hospital readmissions after adjusting for patient age, LOS, race, and diagnosis.


Use of geospatial modeling offers independent predictive value of patients at increased risk of readmission and may allow health care administrators to focus their efforts on these groups to reduce preventable readmissions. This research could transform policy efforts to assign accountability for readmissions and policies to prevent readmissions.

Learning Areas:

Communication and informatics
Conduct evaluation related to programs, research, and other areas of practice

Learning Objectives:
Explain the impact of patient factors associated with hospital readmissions Describe spatial clustering of hospital readmissions Discuss how spatial analysis may improve the ability to elucidate potential target areas for policies to help reduce preventable readmissions

Keyword(s): Hospitals, Geographic Information Systems (GIS)

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: As the lead GIS Analyst on this spatial readmissions project, I have overseen the development and implementation of this work and have worked extensively with statisticians and clinicians to make it practicable. My research interests include translation of complex spatial methodologies.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.