Although the items were hierarchically ordered, it was found that patients did not use the VAS linearly over the full range and that the VAS could at best be considered to contain 10 category groupings. However, this was a small, underpowered study and made certain assumptions about the form of the Rasch model, which would be challenged in modern Rasch analysis protocols. Two other studies have employed the Rasch model to evaluate the VAS response format used in a clinical performance test and a fatigue Fruquintinib severity scale. In both studies the VAS was converted into a 0�C10 Likert scale, which makes assumptions about the scores within each 10 mm step on the scale. The results from these studies showed that categories needed to be combined to achieve fit to the Rasch model. In summary, the VAS continues to be interpreted as an interval scale, rather than a categorical scale as proposed previously and those studies that have used Rasch analysis have investigated scales that used the VAS format, rather than the pain VAS itself. This paper aims to examine the scaling properties and responsiveness of the pain Visual Analogue Scale using Rasch analysis and the implication of the findings for the interpretation of its sensitivity to change along the trait. A strategy was employed whereby the seven repeated VAS pain items across the baseline week, as described above, were treated as though they belonged to a single scale. In other words, the measurement for day one was considered item 1, for day two item 2, and so on. Since the thickness of a cross marked on a VAS may exceed one millimetre, or the interpretation of the exact location may vary by a millimetre, we divided the VAS scores by 2, thus reducing the range of each item to 0�C50 points. We chose not to group the VAS data into 7�C10 categories as proposed by some because we specifically wanted to test if the raw data is indeed an interval scale. Data from the items were fitted to the partial credit Rasch measurement model to determine if the ��scale�� satisfied the expectation of the Rasch model, in other words to examine fit to the model. The Rasch model is a probabilistic model, that expresses the probability of an item that represents a given level of ability being passed by people with a given level of ability, as a TP-0903 inhibitor logistic function of the difference between item difficulty and person ability.