Rasch analysis of the Dutch version of the Oxford elbow score

BACKGROUND The Oxford elbow score (OES) is a patient-rated, 12-item questionnaire that measures quality of life in relation to elbow disorders. This English questionnaire has been proven to be a reliable and valid instrument. Recently, the OES has been translated into Dutch and examined for its reliability, validity, and responsiveness in a group of Dutch patients with elbow pathology. The aim of this study was to analyze the Dutch version of the OES (OES-DV) in combination with Rasch analysis or the one-parameter item response theory to examine the structure of the questionnaire. METHODS The OES-DV was administered to 103 patients (68 female, 35 male). The mean age of the patients was 44.3 ± 14.7 (range 15-75) years. Rasch analysis was performed using the Winsteps(®) Rasch Measurement Version 3.70.1.1 and a rating scale parameterization. RESULTS The person separation index, which is a measure of person reliability, was excellent (2.30). All the items of the OES had a reasonable mean square infit or outfit value between 0.6 and 1.7. The threshold of items were ordered, so the categories can function as intended. Principal component analysis of the residuals partly confirmed the multidimensionality of the English version of the OES. The OES distinguished 3.4 strata, which indicates that about three ranges can be differentiated. CONCLUSION Rasch analysis of the OES-DV showed that the data fit to the stringent Rasch model. The multidimensionality of the English version of the OES was partly confirmed, and the four items of the function and three items of the pain domain were recognized as separate domains. The category rating scale of the OES-DV works well. The OES can distinguish 3.4 strata. This conclusion can only be applied to elbow dislocations, which were the largest group of patients studied.