均衡样本剩余牙本质厚度组间差异的实验探讨

Experimental Study on Reducing Bias of Remaining Dentine Thickness between Groups

【目的】探讨牙本质黏结实验中均衡样本剩余牙本质厚度组间差异的有效方法。【方法】90颗新近拔除的上颌第三磨牙,用低速金刚砂锯于流水下切割牙冠暴露全层牙本质,用螺旋CT以0.625mm精度扫描牙体样本。使用NIH图像分析软件测量髓腔顶到切割面的剩余牙本质厚度D。对D进行排序,然后采用分层随机分组方法,将样本均分为9组,计算各组的均数和标准差,结果进行单因素方差分析及LSD-t检验(α=0.05),所得的结果与使用随机数进行随机分组的结果比较,观察两种分组方法的效果差异。【结果】单因素方差分析结果表明,使用CT图像为依据的各组样本的剩余牙本质厚度组间差异无统计学意义(P=0.998)。LSD-t检验结果表明,各组样本两两比较,剩余牙本质厚度组间差异无统计学意义(P>0.05);而使用随机数进行随机分组的结果虽然也无统计学差异(P=0.352),但是LSD-t检验显示分组间出现统计学差异。【结论】用螺旋CT测量剩余牙本质厚度,再根据厚度分层随机分组,可有效均衡因样本变异或操作误差造成的组间差异,比单纯使用随机数分组所获得的分组组间差异更均衡,效果更理想。在牙本质黏结实验中,可用此法平衡样本剩余牙本质厚度差异所起的作用。

[ Objective ] To find an effective method to reduce bias of remaining dentin thickness between groups during experimental dentin bond test. [Methods] Ninety newly extracted human maxillary third molars were selected. The occlusal third of the molar crowns was removed by means of a water-cooled slow-speed diamond saw. Specimens were scanned using a spiral CT with an accuracy of 0.625 mm. The remaining dentin thickness from the dentin surface to the nearest portion of the pulp chamber was measured by analyzing the CT images with NIH software. Specimens were ranked according to the remaining dentin thickness, and were averagely divided into nine groups by means of stratified randomization. Data of the nine groups were collected and analyzed with SPSS for one-way ANOVA with LSD-t test for multiple comparisons （α=0.05）, data were compare to the way of random number randomization. [ Results ] The result of ANOVA showed that based on the CT image method there was no significant difference in the remaining dentin thickness among the nine groups （P〉 0.05 ）, while no significant difference was also found in the LSD-t test （P〉0.05）, in the way of random number randomization though there was alsono significant difference in the remaining dentin thickness among the nine groups （P〉 0.05）, but some group showed significant difference in the LSD-t test （P〉 0.05）. [ Conclusions ] It was an effective method to reduce bias of remaining dentin thickness between groups which was inevitably brought into because of sample variance and manipulation error by spiral CT scanning and stratified randomization, the effect was more better than the way used of random number randomization. The method used of stratified randomization based on CT image was suggested to balance the effect of remaining dentin thickness during experimental dentin bond test.