Data sources Three electronic databases including Medline, Embase and Cochrane Central were searched for finding the relevant literature. Cross-referencing, reviews and editorials were also used to identify original studies. Unpublished studies/ Grey literature were excluded.

Study selection Both observational and interventional studies were included for the review and only the studies with a minimum of 3 year follow-up period and which have reported the association between the predictor and tooth loss via odds/risk ratio or their derivatives were selected.

Data extraction and synthesis Data like study characteristics, assessed predictors (risk factors or indicators), tooth loss per year as well as the association estimated (as OR, RR or any derivative) were extracted from the included studies. Meta-analysis and meta-regression analysis was carried to find the impact of predictors on the tooth loss.

Results After screening, 20 studies were finally included in the study with a total of 15,422 participants. The mean age of the patient was 44.5 (range: 31-54 years) while the mean follow-up period was 12 years (range: 4-20 years). Predictors were pooled as patient level and tooth level predictors. Strong association was found for most of the patient level (age, compliance, diabetes, IL-1 polymorphism, diabetes and smoking) as well as tooth level predictors (Bone loss, pocket depth, tooth type, furcation involvement and mobility) except for sex (patient level predictor) and endodontic involvement (tooth level predictor).

Conclusions The meta-analysis depicted in the review exhibited strong association for most of the predictors which can be used by the dentists for treatment planning. Overall tooth level factors have stronger associations than patient level factors.

数据来源搜索了包括Medline，Embase和Cochrane Central在内的三个电子数据库，以查找相关文献。交叉引用，评论和社论也用于确定原始研究。未发表的研究/灰色文献被排除在外。

研究选择本研究包括观察性研究和介入性研究，仅选择随访时间最少为3年的研究，这些研究报告了通过风险比或风险比或其预测值报告了预测因素与牙齿脱落之间的关联。

数据提取和综合从纳入的研究中提取数据，例如研究特征，评估的预测指标（风险因素或指标），每年的牙齿脱落以及估计的相关性（以OR，RR或任何派生形式）。进行荟萃分析和荟萃回归分析，以发现预测因素对牙齿脱落的影响。

结果筛选后，最终有20项研究被纳入研究，共有15,422名参与者。患者的平均年龄为44.5（范围：31-54岁），而平均随访期为12年（范围：4-20岁）。将预测指标汇总为患者水平和牙齿水平预测指标。除性别以外，大多数患者水平（年龄，依从性，糖尿病，IL-1多态性，糖尿病和吸烟）以及牙齿水平预测因子（骨丢失，口袋深度，牙齿类型，分叉受累程度和活动性）均具有很强的关联性（患者水平预测指标）和牙髓受累（牙齿水平预测指标）。

结论评价中描述的荟萃分析显示出与大多数预测因素密切相关，牙医可以将其用于治疗计划。总体牙齿水平因素比患者水平因素具有更强的关联性。