فهرست برای کنترل کیفیت در بررسی های آنتروپومتریک
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|4707||2004||4 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : International Journal of Industrial Ergonomics, Volume 34, Issue 6, December 2004, Pages 479–482
In an anthropometric measurement process, there are many factors which are error sources even if the observers are highly trained. Systematic or bias errors are possible and they are not clearly noticeable. In this work we propose two simple and adimensional indices to use in the quality control of anthropometric surveys. Such indices are applicable only to a few anthropometric variables but the results found by this control can be an indication of the quality of the complete measurement process. The indices are defined as
Generally, the accuracy and precision of research experimental devices are obviously known and the approximate true value of measures can be anticipated. Thus, the absolute and relative errors obtained for a measure are representative of the quality of the method and are used as a control of the complete measurement process. On the other hand, statistical variables, e.g. standard deviation, are obtained as information of data variability. These values could be representative of the experimental sample's uniformity since the precision of measurement process is also previously known. Occasionally, they can be used as a measure of the experimental process quality. when an individual value is really out of range, the possible causes are investigated. Only if the existence of a methodological mistake is proven, the value is rejected and the process is repeated correctly. It is a usual and simple quality control of the measuring process and the research methodology. In some studies this is not possible to do. The observer cannot perceive the anomalous measures when the norm has a very wide range. That happens when the size differences among the subjects of a sample are much higher than the accuracy of experimental devices, sometimes a factor of 10 or higher. This is the case of anthropometric researches. In such studies, the direct and immediate quality control of the measurement process is easily missed. To avoid this, many protocols are used to rise the degree of accuracy of the measures (Meunier and Yin, 2000), but there are many factors in human measurement that intervene as sources of error and results can be systematically different in spite of the observers being highly trained (Kemper and Pieters, 1974). Thus, systematic or bias errors are possible and they are not clearly noticeable. Implicitly it is supposed that the standard deviations and variation coefficients are only representative of the population sample variability and that measure errors are negligible. However, occasionally there are systematic errors in the measurement process, which could have a significant effect on both mean values of experimental variables and their standard deviations and could cause mistaken conclusions over different populations. In this work we propose two adimensional indices to use in the quality control of the anthropometric measurement process. The observer must take those measurements again when the index values are higher than a selected threshold before approving or rejecting the value. Such indices are applicable only to a few anthropometric variables but the results found by this control can be an indication of the quality of the complete measurement process.
نتیجه گیری انگلیسی
The variables se and ses (Fig. 2) have statistical distributions that can be fit closely to a normal curve, as usual in this kind of surveys. Only in four cases se values over 7% were obtained and a new measurement process was performed. This fact increased 1.22% the total work, which did not result very expensive. The mean of the relative errors obtained with the quality control used are close to 1%. It seems accurate enough for this kind of anthropometrics data. The se and ses indices are a simple and straightforward but limited method of quality control. These indices are basically a direct warning signal for the variables that define them and an indirect signal for the rest of the variables measured in any study. The databases where the five variables were available have been used. Generally, the individual measures are unpublished and it is not possible to obtain the indices for each subject of the sample. Consequently, we are using the mean values of the corresponding magnitudes to calculate the indices. This way some small differences can be introduced with regard to the mean obtained from individual index values. On the other hand, in Table 1 a great variability amongst the values of the indices calculated for the samples published in the bibliography can be noticed. The extreme values for the samples are 0% (ses, British female) and 36.6% (se, Japanese female), however, this last results is so high that a transliteration error should not be ruled out. The high accuracy of the shoulder–elbow length measures in the sitting position should be emphasised. The index ses is close to 1% in most cases (6 of 7) studied. To ascribe the variability of these results to reliable differences between populations does not seem possible. Thus, measure errors obtained in the standing position are several times higher than obtained in the sitting position. The same landmarks are used in both cases: acromion, olecranon and reference planes. The differences of accuracy between variables controlled in the two positions can be attributed either to a vertical stability increase of the experimental sitting subjects, or to a comfort increase for observers in the measurement process for such position, probably in relation to fix the acromion point. For example, in the standing position, subjects could easily incline to the right or to the left and the measure changes as a result of this movement. Anyhow, the lack of accuracy and precision in anthropometric measures has many causes that have different significance depending on the measurement method. But this is not the main objective of this study that is to propose a method in order to know if the measures have been correctly taken. Nevertheless, it can be deduced that it is possible to reduce the experimental errors approximately in a 1/10 factor at least for the measures in the standing position, through a quality control method like the one we propose in this paper.