Department of Oral Public Health (OPH), Academic Centre for Dentistry Amsterdam (ACTA), University of Amsterdam and Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
Zilveren Kruis Achmea, Zeist, The Netherlands.
PLoS One. 2021 Nov 12;16(11):e0259495. doi: 10.1371/journal.pone.0259495. eCollection 2021.
Various models are available to assess caries risk in individuals. In general past caries experience is considered as the best single predictor for future caries development in populations. Likewise, recent restorations have been used to predict future restorations. We aimed to evaluate a classification model for risk categories for dental caries in children based on claims data from Dutch healthcare insurance company Zilveren Kruis. The baseline caries risk categories were derived from the number of claimed restorations in two baseline years (2010 through 2011). These categories were defined as low (no new restorations), moderate (1 new restoration), and high (2 or more new restorations). First, we analyzed the relationship between baseline caries risk categories and the number of new restorations during 3 years of follow-up (2012 through 2014). Secondly, we used negative binominal two-level analyses to determine the accuracy of our classification model in predicting new restorations during follow-up. Thirdly, we reclassified the participants after 3 years and determined the changes in the categorization. We included insurance claims data for the oral healthcare services in 28,305 children and adolescents from 334 dental practices for the period 2010-2014. At baseline, 68% of the participants were in risk category low, 13% in moderate and 19% in high. The mean number of new restorations during follow-up was 0.81 (SD 1.72) in baseline risk category low, 1.61 (SD 2.35) in moderate, and 2.65 (SD 3.32) in high. The accuracy of the multivariate model for predicting 0/>0 restorations was 50%. After 3 years, 60% of the study participants were in the same risk category, 20% were in a lower, and 21% in a higher risk category. Risk categories based on claimed restorations were related to the number of new restorations in groups. As such, they could support planning and evaluation of oral healthcare services.
有多种模型可用于评估个体的龋齿风险。一般来说,过去的龋齿经历被认为是人群中未来龋齿发展的最佳单一预测指标。同样,最近的修复情况也被用于预测未来的修复情况。我们旨在评估一种基于荷兰医疗保险公司 Zilveren Kruis 理赔数据的儿童龋齿风险分类模型。基线龋齿风险类别是根据两年(2010 年至 2011 年)的理赔数量得出的。这些类别定义为低(无新修复)、中(1 个新修复)和高(2 个或更多新修复)。首先,我们分析了基线龋齿风险类别与三年随访期间(2012 年至 2014 年)新修复数量之间的关系。其次,我们使用负二项式两级分析来确定我们的分类模型在预测随访期间新修复的准确性。第三,我们在三年后重新分类参与者,并确定分类的变化。我们纳入了 2010 年至 2014 年期间 334 家牙科诊所的 28305 名儿童和青少年的口腔保健服务理赔数据。基线时,68%的参与者处于低风险类别,13%处于中风险类别,19%处于高风险类别。在低风险类别中,随访期间新修复的平均数量为 0.81(SD 1.72),中风险类别为 1.61(SD 2.35),高风险类别为 2.65(SD 3.32)。预测 0/>0 个修复的多变量模型的准确性为 50%。三年后,60%的研究参与者处于相同的风险类别,20%处于较低的风险类别,21%处于较高的风险类别。基于理赔的风险类别与群体中新修复的数量相关。因此,它们可以支持口腔保健服务的规划和评估。