D'haene E G, Crombag F J, Tertoolen J F
Br Med J. 1974 Sep 21;3(5933):708-11. doi: 10.1136/bmj.3.5933.708.
Four thyroid function indices were determined in the sera of 181 patients: total thyroxine (T-4), the triiodothyronine resin uptake (T-3 B.C.-index), the free thyroxine (T-4) index, and the effective thyroxine ratio (E.T.R.).A statistically significant correlation between the free T-4 index and E.T.R. could be established though this correlation seemed to be worse than could be expected from the literature. Two sets of correlation coefficients were shown: one obtained with the ordinarily used Pearson or product-moment correlation and one with the Spearman rank correlation. Because the assumptions and requirements for the proper use of the product-moment correlation statistics, such as a normal distribution of both variables, are seldom met, expecially not if a selected number of controls and patients is used for the correlation, it is advisable to use some kind of rank correlation statistics instead.The advantages and disadvantages of the E.T.R. and free T-4 index determination are described.
测定了181例患者血清中的四项甲状腺功能指标:总甲状腺素(T-4)、三碘甲状腺原氨酸树脂摄取率(T-3 B.C.-指数)、游离甲状腺素(T-4)指数和有效甲状腺素比值(E.T.R.)。尽管游离T-4指数与E.T.R.之间的相关性似乎比文献预期的要差,但仍能建立起统计学上的显著相关性。给出了两组相关系数:一组是用常用的皮尔逊或积差相关法得到的,另一组是用斯皮尔曼等级相关法得到的。由于很少能满足积差相关统计正确使用的假设和要求,比如两个变量都呈正态分布,特别是当用于相关性分析的对照组和患者数量有限时,因此建议改用某种等级相关统计方法。文中描述了E.T.R.和游离T-4指数测定的优缺点。