中文摘要：

目的：建立侵袭性牙周炎（aggressive periodontitis,AgP）患者牙周治疗后远期失牙的预测模型。方法：回顾分析85名在北京大学口腔医院牙周科完成牙周治疗并相继于3至11年后复查的AgP患者。获取患者的一般信息、牙周检查数据和影像学资料,用Logistic回归筛选牙周治疗后远期失牙的相关因素,建立预测模型,用受试者工作特征（receiver operator characteristic,ROC）曲线分析预测效力。结果：在积极牙周治疗后的随访期内,85名受试者中有22人失牙,共计55颗。纳入Logistic回归方程的失牙相关危险因素包括初始骨吸收多、根形态异常牙多、探诊出血严重的位点多以及不定期的牙周维护。ROC曲线分析显示预测的敏感性和特异性可同时接近80%。结论：建立的侵袭性牙周炎患者治疗后失牙的预测模型对临床有参考意义。

英文摘要：

Objective：To establish a predictive model for long-term tooth loss of patients with aggressive periodontitis （AgP） after periodontal treatment. Methods： Patients diagnosed as AgP in Department of Periodontology, Peking University School and Hospital of Stomatology, who were re-evaluated 3 to 11 years after periodontal treatment were enrolled （ n = 85 ）. Logistic regression was performed to select background, periodontal and radiographic factors which were related to long-term post-treatment tooth loss. A predictive model was built and analyzed by receiver operator characteristic （ROC） curve. Results： After periodontal treatment, 55 teeth from 22 patients lost further. High prevalence of baseline bone loss, root abnormality, and residual severe bleeding sites, as well as poor compliance to maintenance were detected as risk factors in the predictive model. ROC analysis found the sensitivity and speci- ficity of the model could reach up to 80% simultaneously. Conclusion： Predictive model for post-treatment tooth loss of patients with AgP is an important adjunct in clinical practice.