ObjectiveTo analyze the effectiveness of bone grafting through windowing at the femoral head-neck junction for the treatment of osteonecrosis with the segmental collapse of the femoral head. MethodThe clinical data were retrospectively analyzed from 106 patients (131 hips) with osteonecrosis with the segmental collapse of the femoral head who underwent bone grafting through windowing at the femoral head-neck junction between March 2011 and December 2013. There were 78 males and 28 females, with an average age of 31.3 years (range, 17-43 years). The body mass index ranged from 16.5 to 36.5 (mean, 24.2) . There were 53 cases of corticosteroid-induced osteonecrosis of the femoral head (ONFH), 18 cases of alcohol-induced ONFH, and 35 cases of idiopathic ONFH. According to Association Research Circulation Osseous (ARCO) classification system, 105 hips were rated as stage Ⅲa, and 26 hips as stage Ⅲb; according to the China-Japan Friendship Hospital (CJFH) classification system, 41 hips were classified as C+L1 type, 13 hips as L2 type, and 77 hips as L3 type. Harris score was used for the effectiveness evaluation. The clinical failure cases were defined as patients who need total hip arthroplasty, or had a Harris score of less than 70 points. The Cox risk model analysis and Kaplan-Meier survival curves were used for multivariate analysis and univariate analysis. ResultsThe average follow-up period was 27.9 months (range, 4-51 months). Solid fusions of bone graft were observed at 1.0-1.5 years after operation. The Harris score at last follow-up was 81.41±11.93, showing significant difference when compared with preoperative score (63.24±9.98) (t=13.710, P=0.000) . The results were excellent in 5 hips, good in 41 hips, fair in 57 hips, and poor in 28 hips, with an excellent and good rate of 35.1%. Thirty-three hips were classified as clinical failure. A progressive collapse of the femoral head was observed in 22 hips. The single factor analysis showed that preoperative ARCO stage, preoperative CJFH type, and preoperative Harris hip score were risk factors for clinical failure (P<0.05) . The Cox risk model showed that ARCO stage Ⅲb was independent risk factor for clinical failure (P<0.05) . The Kaplan-Meier survival curves showed that ARCO stage Ⅲa patients had a better effectiveness than ARCO stage Ⅲb patients. ConclusionsBone grafting through windowing at the femoral head-neck junction has a good effectiveness in patients at ARCO stage Ⅲa, while patients at ARCO stage Ⅲb and patients of CJFH types L2 and L3 have high clinical failure rates.