Tiffanie Guillot1, Sylvain Roche2, Pascal Rippert3, Dalil Hamroun4, Jean Iwaz2, René Ecochard2, Carole Vuillerot5. 1. Service de Biostatistique-Bioinformatique, Pôle Santé Publique, Hospices Civils de Lyon, Lyon, France. 2. Service de Biostatistique-Bioinformatique, Pôle Santé Publique, Hospices Civils de Lyon, Lyon, France; Université de Lyon, Lyon, France; Université Lyon 1, Villeurbanne, France; CNRS UMR 5558, Laboratoire de Biométrie et Biologie Évolutive, Équipe Biostatistique Santé, Pierre-Bénite, France. 3. Hospices Civils de Lyon, Pôle Information Médicale, Évaluation, Recherche, Lyon, France. 4. Direction de la Recherche et de l'Innovation, Centre Hospitalo-Universitaire de Montpellier, Montpellier, France. 5. Université de Lyon, Lyon, France; Université Lyon 1, Villeurbanne, France; CNRS UMR 5558, Laboratoire de Biométrie et Biologie Évolutive, Équipe Biostatistique Santé, Pierre-Bénite, France; L'Escale, Service de Médecine Physique et de Réadaptation Pédiatrique, Hôpital Femme-Mère-Enfant, Hospices Civils de Lyon, Bron, France. Electronic address: carole.vuillerot@chu-lyon.fr.
Abstract
OBJECTIVE: To examine whether a Rasch analysis is sufficient to establish the construct validity of the Motor Function Measure (MFM) and discuss whether weighting the MFM item scores would improve the MFM construct validity. DESIGN: Observational cross-sectional multicenter study. SETTING: Twenty-three physical medicine departments, neurology departments, or reference centers for neuromuscular diseases. PARTICIPANTS: Patients (N=911) aged 6 to 60 years with Charcot-Marie-Tooth disease (CMT), facioscapulohumeral dystrophy (FSHD), or myotonic dystrophy type 1 (DM1). INTERVENTIONS: None. MAIN OUTCOME MEASURE(S): Comparison of the goodness-of-fit of the confirmatory factor analysis (CFA) model vs that of a modified multidimensional Rasch model on MFM item scores in each considered disease. RESULTS: The CFA model showed good fit to the data and significantly better goodness of fit than the modified multidimensional Rasch model regardless of the disease (P<.001). Statistically significant differences in item standardized factor loadings were found between DM1, CMT, and FSHD in only 6 of 32 items (items 6, 27, 2, 7, 9 and 17). CONCLUSIONS: For multidimensional scales designed to measure patient abilities in various diseases, a Rasch analysis might not be the most convenient, whereas a CFA is able to establish the scale construct validity and provide weights to adapt the item scores to a specific disease.
OBJECTIVE: To examine whether a Rasch analysis is sufficient to establish the construct validity of the Motor Function Measure (MFM) and discuss whether weighting the MFM item scores would improve the MFM construct validity. DESIGN: Observational cross-sectional multicenter study. SETTING: Twenty-three physical medicine departments, neurology departments, or reference centers for neuromuscular diseases. PARTICIPANTS: Patients (N=911) aged 6 to 60 years with Charcot-Marie-Tooth disease (CMT), facioscapulohumeral dystrophy (FSHD), or myotonic dystrophy type 1 (DM1). INTERVENTIONS: None. MAIN OUTCOME MEASURE(S): Comparison of the goodness-of-fit of the confirmatory factor analysis (CFA) model vs that of a modified multidimensional Rasch model on MFM item scores in each considered disease. RESULTS: The CFA model showed good fit to the data and significantly better goodness of fit than the modified multidimensional Rasch model regardless of the disease (P<.001). Statistically significant differences in item standardized factor loadings were found between DM1, CMT, and FSHD in only 6 of 32 items (items 6, 27, 2, 7, 9 and 17). CONCLUSIONS: For multidimensional scales designed to measure patient abilities in various diseases, a Rasch analysis might not be the most convenient, whereas a CFA is able to establish the scale construct validity and provide weights to adapt the item scores to a specific disease.
Authors: Karlien Mul; Tatiana Hamadeh; Corinne G C Horlings; Rabi Tawil; Jeffrey M Statland; Sabrina Sacconi; Alastair J Corbett; Nicol C Voermans; Catharina G Faber; Baziel G M van Engelen; Ingemar S J Merkies Journal: Eur J Neurol Date: 2021-05-02 Impact factor: 6.089