We reviewed the radiographs of 49 patients with Perthes' disease at the stage of fragmentation and also after the end of skeletal growth to assess the value of the lateral pillar classification of Herring. The average age of the patients at diagnosis was 7 years 6 months and the mean follow-up was 24 years. Ten of the 11 Herring group-A hips showed good reconstruction of the femoral head. There were good results in group-B hips when the patients were less than nine years of age at diagnosis. All 11 group-C patients showed hip deformity at follow-up. The Herring classification provides a valid long-term prognosis in Perthes' disease, although age at diagnosis is also an important prognostic factor. The classification is relatively easy to apply, is reliable, and requires only an anteroposterior radiograph taken during the fragmentation stage of the disease.

We have attempted to identify the most important long-term prognostic factors in Perthes' disease by studying 61 patients affected unilaterally. The average age at diagnosis was 7 years 5 months and at follow-up it was 32 years, an average interval of 25 years. The age at diagnosis, age at follow-up, Catterall group, acetabular coverage, femoral head subluxation and the other head-at-risk signs were statistically correlated with Stulberg, Cooperman and Wallensten (1981) radiographic classes and the Iowa hip score. Statistically significant correlations were found between Stulberg class and Iowa hip score; age at diagnosis and Stulberg class; age at follow-up and Iowa hip score; and between lateral subluxation of the femoral head and Iowa hip score. Three age-groups of patients were found to carry different long-term prognoses. Those below five years of age at diagnosis showed a statistically significant correlation between Catterall group and Stulberg Classes I and II. Patients between five and nine years of age at diagnosis showed a significant correlation between Catterall group and Stulberg Classes I, II, III and IV while in patients diagnosed after nine years of age there was no statistical correlation between Catterall group and Stulberg class, all having a poor prognosis and ending up in Stulberg Classes III, IV and V.