文献速递牙体牙髓病牙周病

口腔疾病进展的可预测性:基于常规临床数据的21-42年纵向研究

Predictability of oral disease progression across the life course:A 21-42-year longitudinal study using routinely collected clinical data.
PubMed 文献 · J Dent · 2026-07-12

该研究分析了198名荷兰军人21-42年的临床与影像学记录,使用LASSO回归构建中长期风险预测模型。长期模型AUC为0.705-0.724,中期模型AUC为0.609-0.676。长期预测主要驱动因素为出生年份、吸烟和军衔,中期为最高龋齿评分和影像学骨丧失进展。

要点

  • 常规临床数据可对口腔疾病进展进行中等程度预测,长期模型优于中期模型。
  • 长期预测中出生年份、吸烟和军衔为一致预测因子。
  • 中期预测中仅拔牙、DMFT增量和两者组合可预测。

核心结论

  • 口腔疾病进展可通过常规收集的临床数据进行中等程度预测,长期模型性能优于中期模型。
  • 预测主要由既往疾病经历和生活方式相关因素驱动,可支持风险分层和针对性预防。

研究设计与样本

  • 分析198名荷兰军人(出生年份1958-1973,以男性为主)的临床和影像学记录,随访超过20年(21-42年)。
  • 长期预测模型使用有限的横断面基线信息,中期预测模型使用更详细的纵向临床病史。
  • 采用LASSO回归筛选关键预测因子,通过500次bootstrapping校正乐观度后计算ROC曲线AUC值。

主要发现

  • 长期模型AUC范围为0.705-0.724,中期模型AUC范围为0.609-0.676。
  • 长期模型中所有结局均具有一定可预测性。
  • 中期模型中仅拔牙(≥1)、DMFT增量(≥1)及其组合可预测。
  • 长期模型中出生年份、吸烟和军衔为一致预测因子;中期模型中最具影响力的是最高龋齿评分和影像学骨丧失进展。

临床怎么用

  • 常规收集的牙科数据可用于中等程度预测口腔疾病进展,支持早期识别高风险个体并制定更个性化的预防护理。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

OBJECTIVES: To assess the predictability of long- and medium-term oral disease progression over 21-42 years using clinical and radiographic data from Dutch military personnel. Outcomes included increments in tooth loss, severe bone loss, DMFT, endodontically treated teeth (ETT), and combined outcomes. METHODS: Records of 198 military personnel with >20 years of follow-up were analyzed. Prediction models were developed for long-term risk assessment using limited cross-sectional baseline information, and for medium-term risk assessment using more detailed longitudinal clinical history. Long-term predictors included demographic variables (year of birth, sex, military rank and smoking) and caries-related variables (DT, MT, FT and a proxy for DT). Medium-term predictors included patient characteristics, dental procedures, and radiographic findings. LASSO regression was applied to identify key predictors. Model performance was assessed using ROC curves with AUC values corrected for optimism by 500-fold bootstrapping. RESULTS: The study population consisted of 198 military personnel (predominantly male) with birth years ranging from 1958 to 1973, representing a broad age range and both military rank categories. Most outcomes showed moderate predictability, with AUC values ranging from 0.705-0.724 for long-term models and 0.609-0.676 for medium-term models. In long-term models, all outcomes were predictable to some extent. In medium-term models, only extractions (≥1), DMFT increment (≥1), and their combination were predictable. Year of birth, smoking, and military rank were consistent predictors in long-term models, whereas highest caries score and radiographic bone loss progression were most influential in medium-term models. CONCLUSIONS: Oral disease progression can be predicted to a moderate extent using routinely collected clinical data, with better performance for long-term than for medium-term models. Prediction is mainly driven by prior disease experience and lifestyle-related factors and may support risk stratification and targeted prevention. CLINICAL SIGNIFICANCE: Routinely collected dental data enable moderate prediction of oral disease progression, supporting earlier identification of high-risk individuals and more tailored preventive care.

来源:J Dent,查看原文。版权归原作者所有。

引用

  1. de la Court J, Opdam N, Laske M, et al. Predictability of oral disease progression across the life course:A 21-42-year longitudinal study using routinely collected clinical data. J Dent. 2026. PMID: 42437596. doi:10.1016/j.jdent.2026.106896

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。