Li A, Yan B, Tan X
Department of Hygiene, Hubei Medical University, Wuhan.
Zhonghua Yu Fang Yi Xue Za Zhi. 1998 Jan;32(1):16-8.
To detect, diagnose and treat for endemic fluorosis earlier.
Six kinds of indices, such as environmental fluoride level, were collected from the population in epidemic and non-epidemic areas of endemic fluorosis with a 1:1 paired-match design. A discriminant analysis model was established by multivariate analysis. Levels of fluoride in environment and biological materials were determined by fluoride electrode method. Living condition of the subjects were measured and interviewed. Function of skeletons and joints was measured. Biochemical and enzyme indices were measured with reagent kits and gel electrophoresis. Other indices were measured by interview. All data collected were analyzed by SAS and MDAS computer software.
There was significant overall difference between four kinds of discriminant functions, with an overall agreement of 85.78% (83.33% to 98.86%), based on resubstitution with sampled data. Posterior probabilities for new classification of sampled data automatically and randomly produced from a computer were 86.39% to 99.99%.
The discriminant functions mentioned above, except for the third one with a too small sample size, can be used in early discrimination of endemic fluorosis caused by exposure to coal burning, or in evaluation for the effectiveness of pharmaceutical therapy, with a power of 95%.
更早地检测、诊断和治疗地方性氟中毒。
采用1:1配对设计，从地方性氟中毒流行区和非流行区人群中收集环境氟含量等6种指标。通过多变量分析建立判别分析模型。采用氟离子电极法测定环境和生物材料中的氟含量。对研究对象的生活状况进行测量和访谈。测量骨骼和关节功能。使用试剂盒和凝胶电泳测定生化和酶指标。其他指标通过访谈进行测量。收集的所有数据采用SAS和MDAS计算机软件进行分析。
基于对抽样数据的重代入，四种判别函数之间存在显著的总体差异，总体一致性为85.78%（83.33%至98.86%）。由计算机自动随机生成的抽样数据新分类的后验概率为86.39%至99.99%。
上述判别函数，除第三个样本量过小外，可用于燃煤型地方性氟中毒的早期判别或药物治疗效果评估，效能为95%。
