Regional variation in the utilization of carotid endarterectomy.

Objective: In different regions of the United States, highly variable rates have been documented for a wide range of procedures, such as prostatectomy and caesarean section. It is generally held that this variation is due to inconsistent physician practice patterns or other nonmedical considerations. Only limited research has been conducted regarding vascular surgical operations. We examined national data on the utilization of carotid endarterectomy (CEA) to determine the extent and diversity of regional variations.

Methods: Medicare discharge data quantified the per capita rate of CEA in 50 states and the District of Columbia in 2003. Multiple metrics pertaining to risk factors, socioeconomic status, access to care, provider density, and local health care capacity were quantified. We performed bivariate analysis, Pearson (PC) or Spearman (SC) correlations, and multiple regression modeling.

Results: In 2003, 83,164 CEAs were performed on 28,767,985 enrollees. CEA rates were 28.9 +/- 7.8 per 10,000 (range, 5.6-44.7 per 10,000). The rate of CEA was highly correlated with the number of heart disease deaths (PC = 0.575, P < .0001), deaths by stroke (PC = 0.504, P = .0002), and percentage of adult smokers in a state (PC = 0.643, P < .0001). These three factors held the strongest association with variation in CEA rates. Statistically, they explained 51% of the variation in total number of CEAs (R(2) = 0.5074, P < .0001). Median annual income (PC = -0.608, P < .0001) and percentage of college degrees (PC = -0.606, P < .0001) displayed inverse relationships to CEA rates. Per capita hospital beds (SC = 0.540, P < .0001) and rural health care clinics (SC = 0.518, P < .0001) exhibited positive correlations. The number of physicians or vascular surgeons did not predict higher utilization of CEA.

Conclusion: The strongest correlations for CEA were three markers associated with atherosclerotic disease: percentage of adult smokers and deaths from heart disease or stroke. Geographic variation in this vascular procedure is chiefly associated with variance in markers of disease prevalence, not physician preference or other nonmedical factors. The increased utilization of carotid stenting, accompanied by the participation of a much wider range of medical specialists, may affect this relationship in the future.