Identification of Clusters of Patients With Similar Baseline Characteristics in the FOCAL-MS2 Study Evaluating Patient’s Autonomy
Author(s)
Svetlana Le Ralle, MSc1, Cecile Donze, PhD2, Geraud Paillot, MSc3, Claude Mekies, PhD4, Mikael Cohen, PhD5, Lucie Brechenmacher, MSc1, Alexandre Civet, MSc1, David Pau, MSc1, Amelie Lambert, MSc1, Guillaume Bourel, MSc1, Catherine Mouzawak, MSc6, Agnès Cimerman, MSc1, Patrick Vermersch, PhD7.
1Roche, Boulogne-Billancourt, France, 2Hôpital saint Philibert, Lille, France, 3Association Aventure Hustive, Saint-Malo, France, 4RAMSAY Clinique des Cèdres, Toulouse, France, 5CHU de Nice Pasteur, Nice, France, 6Structure régionale neuro SEP SYNAPSE, Le Vésinet, France, 7CHU de Lille, Lille, France.
1Roche, Boulogne-Billancourt, France, 2Hôpital saint Philibert, Lille, France, 3Association Aventure Hustive, Saint-Malo, France, 4RAMSAY Clinique des Cèdres, Toulouse, France, 5CHU de Nice Pasteur, Nice, France, 6Structure régionale neuro SEP SYNAPSE, Le Vésinet, France, 7CHU de Lille, Lille, France.
Presentation Documents
OBJECTIVES: MSAS is a new questionnaire that aims at evaluating patient autonomy in multiple sclerosis (MS). This research focuses on identifying clusters ofpatients with similar characteristics at baseline.
METHODS: FOCAL-MS2 is a 1-year longitudinal prospective study (2024-2025) evaluating patient's autonomy. Partitioning Around Medoids (PAM) clustering method was used. Number of clusters Selection/validation: K-Means inertia (Elbow Method), maximizing silhouette Coefficient and Calinski-Harabasz index while minimizing the Dunn index. Stability assessment: bootstrap method (random of 50 samples with replacement and noise added in each sample) and Jaccard index. Descriptive statistics were computed from variables included or not in the clustering step. Missing data were not imputed.
RESULTS: 199 MS patients were included: 74% women, mean age at diagnosis 34.3+/ 9.9 years.6 clusters were identified: silhouette mean score of 0.29, Calinski-Harabasz Index was 61.64 and the Dunn Index was 0.11.For the bootstrap approach, clusters all had a Jaccard similarity statistic >=90% over 50 iterations, which characterizes high stability to resampling with replacement. Clusters description: C1(n=30): Patients aged [50-60] years, on sick leave, with a carer and in a wheelchair, with MS duration between [20-25] years. C2(n=42): Male patients aged [40-50], requiring occasional walking assistance, with MS for [10-15] years. C3(n=42): Patients aged [50-60] with personal activities and requiring occasional walking assistance. Patients with MS duration between [15- 20] years. C4(n=30): Retired patients with caregiver and walking aid at all times. Patients with MS duration >25 years. C5(n=19): Patients under 40 years old without activities without caregiver, without walking aid and MS duration <10 years. C6(n=36): Professionally active women under 40 years without caregiver, without walking aid and MS duration <10 years.
CONCLUSIONS: In our study, the PAM-method allowed us to differentiate 6 clusters.
METHODS: FOCAL-MS2 is a 1-year longitudinal prospective study (2024-2025) evaluating patient's autonomy. Partitioning Around Medoids (PAM) clustering method was used. Number of clusters Selection/validation: K-Means inertia (Elbow Method), maximizing silhouette Coefficient and Calinski-Harabasz index while minimizing the Dunn index. Stability assessment: bootstrap method (random of 50 samples with replacement and noise added in each sample) and Jaccard index. Descriptive statistics were computed from variables included or not in the clustering step. Missing data were not imputed.
RESULTS: 199 MS patients were included: 74% women, mean age at diagnosis 34.3+/ 9.9 years.6 clusters were identified: silhouette mean score of 0.29, Calinski-Harabasz Index was 61.64 and the Dunn Index was 0.11.For the bootstrap approach, clusters all had a Jaccard similarity statistic >=90% over 50 iterations, which characterizes high stability to resampling with replacement. Clusters description: C1(n=30): Patients aged [50-60] years, on sick leave, with a carer and in a wheelchair, with MS duration between [20-25] years. C2(n=42): Male patients aged [40-50], requiring occasional walking assistance, with MS for [10-15] years. C3(n=42): Patients aged [50-60] with personal activities and requiring occasional walking assistance. Patients with MS duration between [15- 20] years. C4(n=30): Retired patients with caregiver and walking aid at all times. Patients with MS duration >25 years. C5(n=19): Patients under 40 years old without activities without caregiver, without walking aid and MS duration <10 years. C6(n=36): Professionally active women under 40 years without caregiver, without walking aid and MS duration <10 years.
CONCLUSIONS: In our study, the PAM-method allowed us to differentiate 6 clusters.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
PCR124
Topic
Methodological & Statistical Research, Patient-Centered Research
Topic Subcategory
Patient-reported Outcomes & Quality of Life Outcomes
Disease
Neurological Disorders