Clustering 5-Year Multidimensional Health Care Trajectory Patterns in Alzheimer's Disease and Related Syndromes (Corrected Proof)

Adeline Gallini (1, 2) , Axel Renoux (1, 2) , Silvan Siep (1) , Aurélie Roustan (3) , Thierry Voisin (1, 2) , Guillaume Escudier (4) , Philippe Muller (5) , Virginie Gardette (1, 2)

1 CHU Toulouse - Centre Hospitalier Universitaire de Toulouse
2 Equipe Vieillissement (CERPOP)
3 CH Montauban - Centre hospitalier de Montauban
4 ARS Occitanie - Agence Régionale de Santé Occitanie
5 IRIT - Institut de recherche en informatique de Toulouse

Objective: After diagnosis of Alzheimer's disease and related syndromes (ADRS), personalized care adapted to each patient's needs is recommended to provide a care plan and start symptomatic treatments according to guidelines. Over the past decade, dedicated structures and care have been implemented in various settings. Equal access to ADRS care, health care providers and services is crucial to ensure potential health benefits for everyone. However, the extent of use of recommended services and favorable health care utilization trajectories (HUT) may vary according to individual and contextual characteristics. The aim of this article was to (1) describe HUT patterns after multidimensional clustering of similar trajectories, (2) assess the proportion of individuals presenting favorable HUTs, and (3) identify factors associated with favorable HUTs. Design: Cohort study. Setting and participants: A cohort of 103,317 people newly diagnosed with ADRS identified in the French health reimbursement system (SNDS) was followed for 5 years with their monthly utilization on 11 health care dimensions. Methods: For 3 age groups (65-74, 75-84, ≥85 years), 15 clusters of patients were identified using partitioning around medoids applied to Levenshtein distances. They were qualitatively assessed by pluridisciplinary experts. Individual and contextual determinants of clusters denoting favorable trajectories were identified using mixed random effects multivariable logistic regression models. Results: Clusters with favorable HUTs denoting slow, progressive trajectories centered on at-home care, represented approximatively 25% of the patients. Determinants of favorable HUTs were mostly individual (age, female gender, absence of certain comorbidities, circumstances of ADRS identification, lower deprivation). Contextual determinants were also identified, in particular accessibility to nurses and nursing homes. Inter-territories variance was small but significant in all age groups (from 0.9% to 1.8%). Conclusion and implications: Favorable HUTs remain the minority and many efforts can still be made to improve HUTs. Qualitative studies could help understanding underlying barriers to favorable HUTs.

Mots clés: Alzheimer's disease; Dementia; cohort; multidimensional health care utilization; patterns; trajectory

Domaines: Informatique

Adeline Gallini, Axel Renoux, Silvan Siep, Aurélie Roustan, Thierry Voisin, et al.. Clustering 5-Year Multidimensional Health Care Trajectory Patterns in Alzheimer's Disease and Related Syndromes (Corrected Proof). Journal of the American Medical Directors Association, 2021, in press, ⟨10.1016/j.jamda.2021.01.085⟩. ⟨hal-03200296⟩