四十年正畸患者龋病风险轨迹的纵向研究
该研究分析710例使用Herbst矫治器患者治疗前、拆除时及随访的全景片,按四个十年段(≤1995、1996-2005、2006-2015、≥2016)比较龋病风险轨迹。结果显示治疗开始时患龋比例从32.9%降至19.5%,治疗期间新发龋从59.8%降至27.4%,现有龋和MFT持续预测新发龋但相对重要性随年代变化,机器学习模型预测效能中等。
要点
- 治疗开始时患龋比例从32.9%降至19.5%,治疗期间新发龋从59.8%降至27.4%。
- 现有龋和MFT持续预测新发龋,但相对重要性随年代变化。
- 低患龋率时代仍需警惕正畸患者龋病风险,需个体化监测。
核心结论
- 龋病患病率/发病率及现有龋和MFT的预测价值随时间变化。
- 机器学习和聚类方法可细化风险轨迹,提示预防需时间自适应模型。
- 尽管总体下降,龋病仍是正畸患者的相关挑战。
研究设计与样本
- 纵向研究分析710例Herbst矫治器患者治疗前、拆除时及随访的全景片。
- 按年代分段:≤1995、1996-2005、2006-2015、≥2016。
- 采用单变量logistic回归、多种监督机器学习模型和k-means聚类。
主要发现
- 治疗开始时患龋比例从32.9%(最早十年段)降至19.5%(最近十年段)。
- 治疗期间新发龋从59.8%降至27.4%。
- 现有龋和MFT持续预测新发龋,相对重要性随年代变化。
- 机器学习模型预测效能中等,性能随时间不同。
- 无监督聚类识别出不同风险特征,特征稳定但患病率随年代变化。
关键数值与推荐
- 结局变量为新发龋;预测变量为年龄、性别、龋(D3/4)和缺失/充填牙(MFT)。
临床怎么用
- 尽管总体龋病下降,正畸患者仍需警惕龋病风险。
- 低患龋率时代识别高风险个体需警惕的个体化监测。
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: This study aimed to characterise caries risk trajectories in orthodontic patients starting treatment from the late 1980s to the present and to evaluate whether the predictive value of risk factors for incident caries experience has changed across decades (≤1995, 1996-2005, 2006-2015, ≥2016). MATERIALS AND METHODS: Panoramic radiographs taken before treatment, at debonding, and at follow-up from 710 patients treated with Herbst appliances were analysed. Outcome variables included incident caries experience; predictor variables were age, sex, caries (D3/4), and missing/filled teeth (MFT). Analyses employed univariate logistic regression, multiple supervised machine learning models, and k-means clustering. RESULTS: The proportion of patients presenting with caries at treatment start declined from 32.9 % in the earliest decade to 19.5 % in the most recent, accompanied by a significant reduction in incident caries during treatment (from 59.8 % to 27.4 %). Present caries and MFT consistently predicted incident caries experience, although their relative importance varied across decades. Machine learning models yielded moderate predictive accuracy, with performance differing over time. Unsupervised clustering identified distinct patient risk profiles that remained stable in character but shifted in prevalence across decades. CONCLUSIONS: Both caries prevalence/incidence and the predictive value of present caries and MFT changed over time. Machine learning and clustering approaches provided refined insights into risk trajectories, emphasising the need for temporally adaptive models in preventive care. CLINICAL SIGNIFICANCE: Despite overall declines, caries remains a relevant challenge in orthodontic patients. Identifying high-risk individuals in an era of a low overall caries prevalence requires vigilant, individualised monitoring.
引用
- Ganss C, Bock NC, Jung K, et al. Caries risk trajectories in a changing oral health landscape. A longitudinal study of orthodontic patients across four decades. J Dent. 2025. PMID: 41205987. doi:10.1016/j.jdent.2025.106216
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。