Measuring the accuracy of different ways to identify the 'at-risk' foot in routine clinical practice

We aimed to identify which individual risk factors best predict foot ulceration in routine clinical practice and whether an integrated clinical tool is a better screening tool for future foot ulceration.

Routinely collected clinical information on foot and general diabetes indicators were recorded on the regional diabetes electronic register. Follow-up data on foot ulceration were collected from the same electronic record, the local multidisciplinary foot clinic and community and hospital podiatry paper records. Data were electronically linked to see which criteria best predicted future foot ulceration.

Foot risk scores were recorded on 3719 patients (44% female, mean age 59±15years) across community and hospital clinics. Overall, 851 (22.9%) had insensitivity to monofilaments, in 629 (17.2%) both pulses were absent and 184 (4.9%) had a prior ulcer. In multivariate analysis, the strongest predictors of foot ulceration were prior ulcer, insulin treatment, absent monofilaments, structural abnormality and proteinuria and retinopathy. The sensitivity of predicting foot ulceration was 52% for prior ulcer, 61% for absent monofilaments, 75% for 'high risk' on an integrated risk score and 91% for high and moderate risk combined. The corresponding specificities were 99, 81, 89 and 61%. Positive likelihood ratio was 52 for prior ulcer and 6.8 for foot risk, with negative likelihood ratios of 0.48 and 0.15, respectively.

Integrated foot risk scores are more sensitive than individual clinical criteria in predicting future foot ulceration and are likely to be better screening tools, where excluding false negative results is of paramount importance.