Validation of a system of foot ulcer classification in diabetes mellitus

Objective The lack of a simple, robust classification of diabetic foot ulcers has critically hampered research into optimum patterns of care. We have therefore attempted validation of the previously published S(AD) SAD system, which is based on grading of ulcer features using simple clinical methods. Research design and methods This was a prospective study in which 300 people with ulcers newly referred to a hospital-based multidisciplinary clinic between 1 January 2000 and 1 July 2002 were classified at the time of their first assessment. If a patient had more than one episode, the last to occur was selected as the index ulcer. If two or more ulcers were registered simultaneously, the one which was regarded as the more significant was chosen. Ulcers were categorized according to area, depth, sepsis, ischaemia and neuropathy. All patients were followed for at least 6 months, or until death if earlier. Outcome criteria used were healed and unhealed (unhealed persisting, unhealed at amputation or death) and were cross-tabulated with different baseline variables. Results Ulcers healed in 209 of the 300 patients (69.7%), while 30.0 (10%) had been resolved by amputation (eight major; 22 minor) and 32 (10.7%) by death. Twenty-nine (9.7%) persisted unhealed. There were significant differences in outcome according to area (chi(2) = 25.9, P < 0.001), depth (chi(2) = 33.8, P < 0.001), sepsis (chi(2) = 13.5, P = 0.004) and arteriopathy (chi(2) = 33.7, P < 0.001), but not to denervation (chi(2) = 5.1, P = 0.16). The strength of these associations was confirmed using Somers d: area (r(s) = -0.24, P < 0.001), depth (r(s) = -0.32, P < 0.001), sepsis (r(s) = -0.15, P < 0.01), arteriopathy (r(s) = -0.30, P < 0.001), denervation (r(s) = -0.10, P = 0.08). Logistic regression analysis using area, depth, sepsis and arteriopathy as independent variables, and those which contributed significantly to the model were area (P = 0.01), depth (P < 0.001) and arteriopathy (P < 0.001). Conclusions These data demonstrate that simple clinical methods can be used to categorize features of individual ulcers, and that area, depth and arteriopathy contribute independently to a model to predict outcome. A system of classification such as this is an essential requirement for the categorization of populations with similar features and similar prognosis, which may then be used as the basis for prospective research into optimal wound management.