H. Carmichael1, A. Moore1, L. Steward1, C. G. Velopulos1 1University Of Colorado Denver,Department Of Surgery,Aurora, CO, USA
Introduction:
The Social Vulnerability Index (SVI) is a composite scale formulated by the Centers for Disease Control to determine resource allocation for natural disasters. It includes 15 variables in four categories of socioeconomic status, household composition/disability, minority status/language, and housing/transportation, and is geocoded as a percentile ranking at the census tract level. Because many of these variables are associated with disparity in access to surgical care, SVI is potentially applicable to assess risk and target populations that are likely to present emergently for disease that could have been treated electively. Because regional variation exists in access to care, future interventions depend on understanding disparity at the discrete, local level. We applied the SVI to compare cholecystectomy patients presenting emergently versus electively.
Methods:
We identified patients who had undergone cholecystectomy at our academic medical center over a 6-month period. We excluded patients <18 yo and pregnant patients. Cases were classified as emergent or elective; cases where the patient presented electively for interval operation after a presentation in the emergent setting requiring intervention were excluded. We abstracted patient demographics, residential address, insurance status, chronic and acute symptom duration, diagnosis, and operative outcomes from the EMR. Patient addresses were geocoded to identify their census tract of residence and estimated SVI. Wilcoxon rank sum tests were used for univariable analysis, followed by multivariable logistic regression modeling.
Results:
Of 289 patients who underwent cholecystectomy, 267 met inclusion criteria. Most patients (n=196, 73.4%) had surgery in the emergent setting. Emergent patients lived in areas of greater social vulnerability compared to elective patients (median SVI 75th vs. 64th percentile, p=0.007). On multivariable analysis adjusting for patient age, sex and chronicity of symptoms, having high SVI (>80th percentile) was associated with higher odds of undergoing an emergent versus an elective procedure (OR 2.19, p=0.02). Models were then compared, with AUC of 0.819 for a model including insurance, PCP, minority, and need for interpreter versus AUC of 0.831 for the model using SVI only.
Conclusion:
The SVI has potential utility for examining health care disparities, performing comparably to a more complex model. Because it is a composite measure geocoded at the census tract level for all communities in the United States, it has potential for targeting relatively discrete geographic areas for intervention. Being a geocoded measure also offers opportunity for linking with other datasets using Geographic Information Systems.
