Evaluation of the subchondral fracture in predicting the extent of femoral head necrosis in Perthes disease: a prospective study of 92 patients.

The aim of this study was to evaluate the subchondral fracture as a predictor for the extent of femoral head necrosis in Perthes disease. Out of 392 patients, 92 (23.5%) had a detectable subchondral fracture at the time of diagnosis. There was concordance between predicted Catterall groups on the basis of the extent of the subchondral fracture and the actual Catterall groups at the time of maximal resorption in 61% of the cases, when assessed by an experienced observer. When using the extent of the subchondral fracture to predict Salter-Thompson groups, this observer obtained 89% concordance with the actual Salter-Thompson groups at the time of maximal resorption. The inter-observer agreement between the experienced and a less experienced observer regarding the presence or absence of a subchondral fracture was moderate (weighted kappa 0.59, 87% agreement). When using the extent of the subchondral fracture as a measure of femoral head involvement (Catterall groups), the inter-observer agreement was moderate (weighted kappa 0.46). Patients with detectable subchondral fracture were significantly older (mean 6.5 years) at the time of diagnosis than those without visible fracture (mean 5.2 years). The delay in diagnosis was significantly shorter in the group with subchondral fracture (mean 3.2 months) than among patients without visible fracture (mean 4.9 months). There was no significant difference with regard to sex, pain level, pain localization, or limping gait between the two groups. We conclude that the subchondral fracture is a relatively rare early sign in Perthes disease. When present, it is a useful sign when assessed by an experienced observer as its extent was in fairly good concordance with the extent of femoral head involvement at the time of maximal resorption. Awareness of this radiographic sign will aid the orthopaedic surgeon to establish diagnosis and, to some degree, to predict prognosis early in the course of the disease.