Rasch analysis of the Italian Lower Extremity Functional Scale: insights on dimensionality and suggestions for an improved 15-item version

Abstract

Objective: To investigate dimensionality and the measurement properties of the Italian Lower Extremity Functional Scale using both classical test theory and Rasch analysis methods, and to provide insights for an improved version of the questionnaire. Design: Rasch analysis of individual patient data. Setting: Rehabilitation centre. Participants: A total of 135 patients with musculoskeletal diseases of the lower limb. Results: Patients were assessed with the Lower Extremity Functional Scale before and after the rehabilitation. Rasch analysis showed some problems related to rating scale category functioning, items fit, and items redundancy. After an iterative process, which resulted in the reduction of rating scale categories from 5 to 4, and in the deletion of 5 items, the psychometric properties of the Italian Lower Extremity Functional Scale improved. The retained 15 items with a 4-level response format fitted the Rasch model (internal construct validity), and demonstrated unidimensionality and good reliability indices (person-separation reliability 0.92; Cronbach’s alpha 0.94). Then, the analysis showed differential item functioning for six of the retained items. The sensitivity to change of the Italian 15-item Lower Extremity Functional Scale was nearly equal to the one of the original version (effect size: 0.93 and 0.98; standardized response mean: 1.20 and 1.28, respectively for the 15-item and 20-item versions). Conclusion: The Italian Lower Extremity Functional Scale had unsatisfactory measurement properties. However, removing five items and simplifying the scoring from 5 to 4 levels resulted in a more valid measure with good reliability and sensitivity to change.