Address correspondence to Stephen G. Jones, M.S., BlueCross BlueShield of Tennessee, Medical Informatics Department, 1 Cameron Hill Circle, Bldg. 2.1, Chattanooga, TN 37402; e-mail: firstname.lastname@example.org. Avery J. Ashby, M.S., Soyal R. Momin, M.S., M.B.A., and Allen Naidoo, Ph.D., are with the BlueCross BlueShield of Tennessee, Medical Informatics Department, Chattanooga, TN.
Spatial Implications Associated with Using Euclidean Distance Measurements and Geographic Centroid Imputation in Health Care Research
Article first published online: 24 SEP 2009
© Health Research and Educational Trust
Health Services Research
Volume 45, Issue 1, pages 316–327, February 2010
How to Cite
Jones, S. G., Ashby, A. J., Momin, S. R. and Naidoo, A. (2010), Spatial Implications Associated with Using Euclidean Distance Measurements and Geographic Centroid Imputation in Health Care Research. Health Services Research, 45: 316–327. doi: 10.1111/j.1475-6773.2009.01044.x
- Issue published online: 8 JAN 2010
- Article first published online: 24 SEP 2009
- zip-code centroid;
Objective. To determine the effect of using Euclidean measurements and zip-code centroid geo-imputation versus more precise spatial analytical techniques in health care research.
Data Sources. Commercially insured members from a southeastern managed care organization.
Study Design. Distance from admitting inpatient facility to member's home and zip-code centroid (geographic placement) was compared using Euclidean straight-line and shortest-path drive distances (measurement technique).
Data Collection. Administrative claims from October 2005 to September 2006.
Principal Findings. Measurement technique had a greater impact on distance values compared with geographic placement. Drive distance from the geocoded address was highly correlated (r=0.99) with the Euclidean distance from the zip-code centroid.
Conclusions. Actual differences were relatively small. Researchers without capabilities to produce drive distance measurements and/or address geocoding techniques could rely on simple linear regressions to estimate correction factors with a high degree of confidence.