Dependence clusters in Alzheimer's disease and Medicare expenditures: A longitudinal analysis from the Predictors Study: Health services research / Cost of care. (7th December 2020)

• Record Type: Journal Article
• Authors: Zhu, Carolyn W; Lee, Seonjoo; Ornstein, Katherine; Cosentino, Stephanie; Gu, Yian; Andrews, Howard; Stern, Yaakov
• Is Part Of: Alzheimer's & dementia. Volume 16(2020)Supplement 10
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-12-07
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.041558
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Dependence has been proposed as a holistic, transparent, and meaningful representation of dementia disease severity to quantify and stage disease progression. Modeling clusters in dependence trajectories can help understand changes over time in the course of dementia and related cost of care. Method: 199 initially community‐living patients with probable AD in early stages of the disease were recruited to the Predictors 2 Study from three academic medical centers in the US. Patients were followed for up to 10 years and had at least two Dependence Scale (DS) recorded. Non‐parametric K‐means cluster analysis for longitudinal data (KmL) was used to identify dependence clusters. Medicare expenditures (1999‐2010) were compared between clusters at 6‐month intervals and cumulatively over 5 years. Result: KmL identified two distinct DS clusters: (A) high initial dependence, faster decline, and (B) low initial dependence, slower decline. At baseline, DS score in cluster A was on average 2.25 points higher than in cluster B (p<0.0001). DS scores in both clusters steadily worsened (cluster A by 0.74±0.038 points, cluster B by 0.42±0.038 points every 6 month, both p<0.001), with significant between‐cluster differential rate of change over time (p<0.0001). At baseline, patients in cluster A were older, more likely to be female, have worse functioning and cognition, and more likely to have psychiatric symptoms than those in cluster B. Average per person Medicare expenditures in a 6‐month interval was higher in cluster A than B ($5, 456 vs. $2, 715). Adjusting for patient characteristics, cluster A was associated with an average of $1, 554 higher Medicare expenditures every 6 months cluster B. By year 5, differences in expenditures between clusters accumulated to $54, 557 per person in cluster A, twice as high as $27, 153 for cluster B. Conclusion: Distinct DS clusters with subtle differences at baseline and rates of decline over time were associated with widening difference in expenditures over time. Dependence is a useful measure to capture long‐term costs associated with dementia. Better characterization of dependence clusters has significant implications for understanding disease progression, trial design and care planning.