Related Experiment Video
Updated: Jul 13, 2026
Dynamic Quantitative Sensory Testing to Characterize Central Pain Processing
Published on: February 16, 2017
Rasch Analysis of the Composite Autonomic Symptom Score-31 (COMPASS-31) Using Data From Chronic Pain Patients
Wim Tuinebreijer1, Henk Koning2
1Surgery, University Medical Center Rotterdam, Rotterdam, NLD.
Background:
Patients with pain experience autonomic symptoms that are assessed using the patient-reported outcome measure known as the Composite Autonomic Symptom Score-31 (COMPASS-31).
Objectives:
This study aimed to analyze the Dutch version of the COMPASS-31 using Rasch analysis.
Methods:
Sixty-two patients completed the Dutch version of the COMPASS-31. The sample population comprised 62 patients (35 female; 27 male), with a mean age of 56.9 years (SD=13.7). Rasch analysis was conducted using Winsteps® Rasch Measurement Version 5.6.0 (Winsteps, Beaverton, Oregon, USA).
Results:
The COMPASS-31 domains performed moderately in a group of patients with pain. Infit and outfit indices of the domains were reasonable. Person reliability is below 0.8, the minimum level required for serious decision-making. Item reliability was moderate, except for the gastrointestinal domain, which was good. Only the orthostatic, gastrointestinal, and pupillomotor domains can differentiate between the two patient groups. Not all items within the domains measure a single unidimensional variable, one of the prerequisites of a Rasch model. The probability curve of the different domains worked moderately well. The patients could discriminate among four item levels.
Conclusion:
Our data did not align with the Rasch model. The analysis failed to confirm the scale's reliability, as the domains could differentiate only one or two levels of autonomic symptoms among patients with chronic pain. Person reliability falls below 0.8, the minimum threshold required for meaningful decision-making. Only the orthostatic, gastrointestinal, and pupillomotor domains can distinguish between the two patient groups, while the others identify only one level. This model misfit cannot be attributed solely to the COMPASS-31 instrument but also to the Rasch model assumptions that were not met, such as a small sample size and items with a skip-pattern design, which violate the assumption of local independence.
