[Psychometric Evaluation of the 'German Neurological Fatigue Index for Multiple Sclerosis (NFI-MS-G)' in a Sample of Rehabilitation Patients with Multiple Sclerosis].

Purpose: The purpose of this study was to provide a patient-reported outcome measure for people with multiple sclerosis (MS) comprehensively reflecting the construct of fatigue and developed upon the assumptions of the Rasch model. The Neurological Fatigue Index - Multiple Sclerosis (NFI-MS) is based on both a medical and patient-described symptom framework of fatigue and has been validated. Therefore, in this study the German version of the NFI-MS (NFI-MS-G) consisting of a physical and cognitive subscale and a summary scale was validated.

Method: In this bi-centre-study, 309 people with MS undergoing outpatient rehabilitation or being≥2 months before or after their inpatient rehabilitation completed the German NFI-MS-G twice within 14-21 days together with other questionnaires. Correlation with established questionnaires and Rasch analysis were used for its validation. Additionally, psychometric properties of known-groups validity, internal consistency, test-retest reliability, measurement precision and readability were tested. Finally, the English NFI-MS and German NFI-MS-G were compared with each other to equate the language versions.

Results: The NFI-MS-G showed good internal construct validity, convergent and known-groups validity and internal consistency (Cronbach's alpha 0.84-0.93). The physical subscale showed minor local dependencies between items 1 and 7, 2 and 3 and 4 to 6, that could be treated by combining the respective items to testlets. Unidimensionality was found for the physical and cognitive subscales but not for the summary scale. Replacing the summary scale, a 2-domains subtest measuring the higher-order construct of fatigue was created. Good test-retest reliability (Lin's concordance correlation coefficient of 0.86-0.90) and low floor and ceiling effects were demonstrated. The NFI-MS-G was found easily readable and invariant across groups of gender, age, disease duration, timepoint and centre.

Conclusion: The German version of the NFI-MS comprehensively represents the construct of fatigue and has adequate psychometric properties. The German version differs from the English original version with respect to a lack of unidimensionality of the summary scale and minor local dependencies of the physical subscale that could be canceled out using a testlet analysis.