聚类分析法在碘缺乏病监测数据分析中的应用

Appfication of clustering analysis of data in iodine deficiency disorders surveillance

目的探索聚类分析在碘缺乏病监测数据分析中的应用。方法采用聚类分析方法，对2005年全国碘缺乏病监测结果按合格碘盐食用率、碘盐覆盖率、盐碘中位数、尿碘中位数、甲状腺触诊肿大率、甲状腺B超肿大率、儿童智商指标，应用类平均法进行系统聚类分析，并绘制聚类图。结果31个省、市（自治区）和新疆生产建设兵团成功地被分成3类：西藏为Ⅰ类地区；广东、青海、海南、新疆为Ⅱ类地区；其他为Ⅲ类地区。Ⅰ类地区的合格碘盐食用率、碘盐覆盖率、尿碘中位数分别为21．00％、33．20％、96．70μg／L．Ⅱ类地区分别为71．95％、78．50％、145．40μg／L，Ⅲ类地区分别为93．20％、97．70％、245．20μg／L，Ⅰ、Ⅱ类地区的上述各指标与Ⅲ类地区比较，差异均有统计学意义（P〈0．05）；Ⅰ类地区甲状腺触诊肿大率为11．50％；Ⅰ、Ⅱ、Ⅲ类地区的智商分别为77．00、95．95±4．46、104．59±6．77，任意两类间比较差异有统计学意义（P〈0．05）。结论通过聚类分析方法能够将多个省份按碘缺乏病的防治情况进行分类，Ⅰ类地区的防治状况值得政府重点关注，Ⅱ类地区的防治工作需要加大力度．Ⅲ类地区的防治工作继续保持。聚类分类方法获取的结果，可能对卫生管理部门的决策提供理论依据。

Objective To explore the application of clustering analysis in the study for iodine deficiency disorders（IDD） surveillance. Methods System cluster method of average linkage was applied to study the data obtained from national IDD surveilance in 2005 and a clustering analysis＇ tree chart was drawn. Such data as the rates of consuming qualified iodine salt, coverage rate of iodine salt, median of iodine salt, median urinary iodine, goiter rate detemined by palpation and B-ultrasonography, Child intelligence quotient were investigated. Results Thirty-one provinces or autonomous regions including the Xinjiang Production and Construction Corps had beed assorted into three types. Tibet was claasified into the type Ⅰ ; Guangdong, Qinghai, Hainan, Sinkiang were assorted into the type Ⅱ , other areas were assorted into the type Ⅲ. The rates of consuming qualified iodine salt, coverage rate of iodine salt and median urinary iodine in the type Ⅰarea were 21.00%, 33.20% and 96.70 μg/L, while they were 71.95%, 78.50% and 145.40μg/L, respectively, all being lower than that of typeⅢ （93.20%, 97.70% and 245.20 μ4g/L） with statistical significance （P 〈 0.05）. 11.50% of the children in the type Ⅰ area were detected by palpation to have goiter; the child intelligence quotient in three types of areas, was 77.00, 95.95 ± 4.46 and 104.59 ± 6.77, the differences being statistical significant between any two types（P 〈 0.05）. Conclusion Multiple areas of IDD could be assorted by clustering analysis into distinct types with different epidemic value. More attentions of the government must be paid to the type Ⅱ areas, the prevention and cure work must be enhanced in the second-class areas while it is carried on in the type Ⅲ areas. The results are helpful for decision making for epidemic supervisors in the government.