Information about the 2019 Novel Coronavirus
American College of Surgeons Level I Trauma Centers (ACSL1TCs) meet the same personnel and structural requirements but serve different populations. We hypothesized that these nuanced differences may amenable to description through mathematical clustering methodology.The National Trauma Data Bank 2014 was used to derive information on ACSL1TCs. Explorative cluster hypothesis generation was performed using Ward's linkage to determine expected number of clusters based on patient and injury characteristics. Subsequent k-means clustering was applied for analysis. Comparison between clusters was performed using the Kruskal-Wallis or chi-square test.In 2014, 113 ACSL1TCs admitted 267,808 patients (median = 2220 patients, range: 928-6643 patients). Three clusters emerged. Cluster I centers (n = 53, 47%) were more likely to admit older, Caucasian patients who suffered from falls (P < 0.05) and had higher proportions of private (31%) and Medicare payers (29%) (P = 0.001). Cluster II centers (n = 18, 16%) were more likely to admit younger, minority males who suffered from penetrating trauma (P < 0.05) and had higher proportions of Medicaid (24%) or self-pay patients (19%) (P = 0.001). Cluster III centers (n = 42, 37%) were similar to cluster I with respect to racial demographic and payer status but resembled cluster II centers with respect to injury patterns (P < 0.05).Our analysis identified three unique, mathematically definable clusters of ACSL1TCs serving three broadly different patient populations. Understanding these mathematically definable clusters should have utility when assessing an institution's financial risk profile, directing prevention and outreach programs, and performing needs and resource assessments. Ultimately, clustering allows for more meaningful direct comparisons between phenotypically similar trauma centers.
View details for PubMedID 28688640