Fig. 5.

Statistics decision tree identifying groups of patients showing nodes, F-statistic values, freedom degrees, and p-values. For the TAMBA, the optimum classification value was 26.3°. Patients with a TAMBA angle value lower than or equal to 26.3 were in node 1 (36 feet). This node showed that 77.8% of the patients (28 out of 36) had not been operated on. In node 2 were patients with TAMBA greater than 26.3 (60% were operated, 6/10). Since lower TAMBA values had better follow-up results, we could say that non-operated patients had a higher probability of obtaining a follow-up result lower than 26.3° (F_{1, 44} = 8.708, p = 0.005). The overall correct percentage of classification was 73.9%. For the TCA, the optimum classification value was 39.7°. 75.7% of the patients (28/37) who showed a TCA equal to or lower than 39.7° (node 1) were not operated (37 feet). Node 2 included feet with a TCA greater than 39.7°, of which 55.6% (5/9) were operated. The lower the TCA, the better the results, so non-operated feet had the highest probability of getting the TCA equal to or lower than 39.7° (F_{1, 44} = 4.548, p = 0.039). The overall correct percentage of classification was 71.7%. Regarding plantar flexion of the ankle, the best value that classified this variable was 36°. Nodes 1 and 2 correctly classify 100% of cases. All feet that were surgically treated belonged to the group with 36° or less. Otherwise, when the plantar flexion of the ankle is greater than 36°, all feet were not operated. The higher the plantar flexion of the ankle, the better the results, so the best classified cases were the non-operated feet (F_{1, 44} = 166.160, p ≤ 0.001). The overall correct percentage of classification was 100%