Voss L D, Bailey B J, Cumming K, Wilkin T J, Betts P R
Endocrine Section, Southampton General Hospital.
Arch Dis Child. 1990 Dec;65(12):1340-4. doi: 10.1136/adc.65.12.1340.
The two major components of reliability are accuracy and reproducibility. Three studies of the reliability of height measurement in children are reported. In the first, a standard metre rod was used to spot check the accuracy of installation of 230 measuring instruments in one health district in Wessex, UK. The readings obtained ranged from 90.0 to 108.5 cm and showed the urgent need for the positioning of instruments to be regularly checked. In a second study, to examine the reproducibility of height measurement, two experienced observers measured 10 young children (106.0 to 152.0 cm), three times on five instruments of different design. The observations were blind and in random order. The estimated standard deviation for a single height measurement was generally in the range 0.2-0.3 cm. Over 95% of the variance was attributable to the child, very little to the instrument or observer. Finally, the conditions of the second study were modified to examine the effect on reproducibility of non-blind and non-randomised measurements, as usually occurs in the clinic. A lower but inevitably false estimate of the error was obtained. It is recommended that the error of height measurement, appropriately established and expressed in simple terms, be stated in every published growth study.
可靠性的两个主要组成部分是准确性和可重复性。本文报告了三项关于儿童身高测量可靠性的研究。第一项研究中，使用标准米尺对英国韦塞克斯郡一个健康区的230台测量仪器的安装准确性进行了抽查。所获得的读数范围为90.0至108.5厘米，这表明迫切需要定期检查仪器的定位情况。在第二项研究中，为了检验身高测量的可重复性，两名经验丰富的观察者对10名幼儿（身高在106.0至152.0厘米之间）在五种不同设计的仪器上进行了三次测量。观察是盲法且随机进行的。单次身高测量的估计标准差一般在0.2 - 0.3厘米范围内。超过95%的差异归因于儿童，而归因于仪器或观察者的差异非常小。最后，对第二项研究的条件进行了修改，以检验非盲法和非随机测量（如临床中通常出现的情况）对可重复性的影响。得到了一个较低但必然是错误的误差估计值。建议在每一项发表的生长发育研究中，都应说明经过适当确定并以简单方式表述的身高测量误差。
