Growth prediction in Class III patients using cluster and discriminant function analysis

    loading  Checking for direct PDF access through Ovid


This longitudinal retrospective cephalometric study was undertaken in an attempt to identify subgroups of subjects with Class III malocclusions and to find discriminant functions which would help to differentiate between favourable and unfavourable growers. The material consisted of cephalometric films of 115 Class III untreated patients (59 females and 56 males, with a mean age of 11.6 ± 1.7 and 12.7 ± 1.3 years, respectively) who were observed for a minimum period of 1 year. All subjects were Caucasian and none could achieve an edge to edge occlusion. Hierarchical cluster analysis was used to identify Class III subgroups. Discriminant function analysis (DFA) was first applied to the whole sample and later to each of the clusters produced. Good and poor growers were identified on the basis of the change in Wits measurements with projection on the maxillary/mandibular planes bisector. The cut-off point between good and bad growers was a Wits value of 2.5 mm which was the upper limit of the 95 per cent confidence interval of measurement reproducibility.Three clinically distinguishable clusters were produced, namely long, short and intermediate facial types. The discrimination percentage (80 per cent) achieved when the DFA was performed on the whole sample was satisfactory. However, when the analysis was used on each of the clusters separately, the equation successfully predicted a good or poor outcome in 92 per cent of cluster I, in 85 per cent of cluster II and in 100 per cent of cluster III.

    loading  Loading Related Articles