Economic Evaluation of Socially Assistive Devices among the Institutionalized Elderly with Dementia in the US

Wednesday, June 15, 2016: 8:50 AM
401 (Fisher-Bennett Hall)

Author(s): Byung-Kwang Yoo; Tomoko Sasaki; Shuichi Nishio; Hidenobu Sumioka; Hiroshi Ishiguro

Discussant: Yoko Ibuka, PhD

Abstract

Background: Because of the growing number of dementia elderly with behavioral problems (e.g., aggressive behaviors and agitation) and a high risk of restraint-use for these behavioral problems, the antipsychotic medications utilization have increased sharply – consequently worsening insurers’ financial burdens as well as the overall health outcomes of elderly using these medications. 

Objectives: To measure economic efficiency of two socially-assistive devices “Hugvie” (a human-shaped cushion functioning as a communication medium) and “Telenoid,” (a remotely-operated android designed to appear and behave like a minimalistic human).  These devices are expected to reduce problematic behaviors among elderly with dementia in long-term care facilities.  Both of these devices were developed by a prominent robot scientist (Prof. Hiroshi Ishiguro; this abstract’s coauthor).  The short-term effectiveness of these devices, regarding (a) improvement in emotion and utility (measured in terms of quality-adjusted life year, (QALY)) and (b) reduction in behavioral problems, is suggested in the past demonstrations in Japan and Denmark.

Methods:  We assumed a hypothetical cohort of dementia residents in U.S. long-term care facilities participate in three conversation sessions over a course of three weeks.  In each session, the participants are expected to have conversations with a healthcare staff member using either Hugvie or Telenoid.  We developed two separate standard decision models that can capture the benefits of improvement in behavioral problems: Model 1 focuses on the reduction in cost due to reduction in formal healthcare staff time; and Model 2 focuses on the reduction in medication cost (e.g. antipsychotics) used for the purpose of managing behavior, assuming that staff time and medications are perfect substitutes.  Model parameters are based on the literature and our original studies.  When these models estimate the benefit-cost ratios (BCRs) to be lower than 1 (not cost-saving), we run cost-effectiveness analysis (CEA) to estimate the incremental cost-effectiveness ratio (ICER) of an intervention compared to “doing-nothing.”   

Results: According to the preliminary analyses, the cost reduction is calculated to be $13.03 in Model 1 and $3.75 in Model 2 per resident for a 3-week intervention, while the cost of the intervention is calculated to be $38.52.  These values indicate that the even a lower cost intervention using Hugvie ($60 per unit) is not cost-saving.  However, our break-even analysis showed that the intervention of Hugvie can be cost-saving if the duration of the positive effect is sufficiently long (e.g., at least 62 days).  The probabilistic CEA for Hugvie estimated an ICER mean of $ 49,300 with a relatively narrow 95% confidence interval ($23,000, $119,000) QALY - implying that the Hugvie intervention is mostly cost-effective compared to the commonly-used threshold of $100,000 per QALY.  Other CEAs estimated that Telenoid ($500 for a shared use) needs to be effect 355 days and 30 days to achieve cost-saving and cost-effective, respectively.

Conclusion: Our analysis shows that the Hugvie is likely to be cost-effective and could be cost-saving when the positive effect remains for a long term.  Additional economic evaluation results based on updated parameter data from ongoing interventions will be provided at the time of the presentation.