There have been discrepancies to difficulties to assess a subjective experience such as mood or pain. The difficulties will remain constant from one measurement interval to the next, and especially from one person to another. As for within-subject comparisons, VAS was suited to experimental designs, and Ohnhaus and Adler (1975) and Joyce et al. (1975) found the VAS-pain were more sensitive than the other verbal rating scales (ex. 5-point scales). However, its suitability for between-subject comparisons has been questioned for the reasons that the equivalent positioning of marks on lines by different people does not necessarily convey that they experienced the same pain.
Although VAS provide easily the numerical values for statistical analysis, there have been some problems underlying the use of VAS as a sophisticated statistical technique.
There have been two ways to treat the length of the line as the evaluation of pain intensity in statistical analysis. The one equates the length of the line produced by the subject's mark with the estimation of pain as though the subjects were able to effect one-to-one correspondence between VAS and the pain intensity (parametric or cardinal). The other prefers to treat as the data so obtained non-parametrically (ordinal).
It should be noted that subjects did not necessarily attach the same meanings to the same absolute scores.
Subjective estimates like the VAS do not qualify as interval level scores and are thus not suited for the operations of addition and subtraction and calculations of averages. (Chapman 1976)
The use of rank based non-parametric statistical procedures helps to overcome some of the difficulties associated betwwn-subject comparisons with the VAS, as procedures are based on the assumptions that the data cannot be classified into particular distributions and that its scaling can be represented by ranks.
VAS pain scores obtained in our patients with Herpes Zoster or RSD were treated as ordinal scales and statistical analyses were carried out by use of non-parametric method (Wilcoxon signed rank test, Kruskal Wallis test or Friedman 2 way-ANOVA).
These results have been presented.
抄録全体を表示