After adjusting for age, gender, the number of hospitalizations during the 12-month period prior to surgery, comorbidity, region of residence, location and teaching status of hospital, and race by diabetes interaction, black patients were more likely to have above knee amputation (OR=1.83; 95% confidence interval=1.73 to 1.94) and toe and/or foot amputation (OR=1.37; 95% confidence interval=1.29 to 1.46). Black patients were also less likely to have limb-saving procedures; the OR for percutaneous transluminal angioplasty=0.39 (95% confidence interval =0.36 to 0.43) and OR for lower-extremity bypass=0.62 (95% confidence interval =0.59 to 0.65).

In the fully adjusted models, patients’ region of residence, the location of the hospital in which they received care, and the teaching status of the hospital all influenced the likelihood of undergoing the surgical procedures for both black and white patients. Thus, these factors did not explain the racial difference.

Alternate explanations offered include: unmeasured factors related to severity of illness (such as a delay in presentation following initial manifestations of disease, lack of access to outpatient care, less optimal outpatient care, or patient noncompliance); difference in the expertise of the treating hospitals (although unlikely because the availability of necessary technologies would likely be associated with teaching hospital status, which was controlled in this analysis); patient preferences; and biased decision making.

“The influence of patient preferences could explain our results only if a disproportionate number of black compared with white patients underwent amputation following a failed…” prior procedure (which patients might want to avoid repeating, such as a failed lower extremity bypass).

As an alternative to biased decision-making, the authors posit that the greater likelihood of limb sparing therapy among white patients might be partially explained by the use of this therapy for situations when medical therapy is appropriate. “Whether this is a viable explanation cannot be known until we collect more detailed clinical information as well as information on physician decision making processes.”