Hospice Coding and Data Conventions Contribute to Oversimplify the Presence of Dementia in Hospice.

OBJECTIVES: To document how dementia diagnoses appear in hospice claims, and how these different presentations reflect different hospice utilization. The reliance in the hospice literature on primary diagnosis, in addition to the focus on decedents, may underestimate the true presence of dementia in hospice, and little is known about the health care utilization of hospice patients with dementia as a secondary or not present diagnosis. DESIGN: Secondary data analysis of Medicare claims. SETTING AND PARTICIPANTS: Medicare beneficiaries with 2 or more dementia diagnoses from 2016 to 2018 electing hospice in 2018. METHODS: Beneficiaries were classified based on the presence and position of dementia on their subset of hospice claims: primary diagnosis, secondary diagnosis, and not present. We then compared the demographics and utilization of the 3 claim-based categories of dementia beneficiaries in hospice in 2018. RESULTS: Fewer than half of beneficiaries with a dementia diagnosis in all of their Medicare claims have dementia indicated as the primary diagnosis associated with their hospice claims, and 30% of beneficiaries did not have their diagnosed dementia appear at all on their hospice claims. Hospice length of stay and other utilization characteristics varied markedly across the 3 claim-based categories of dementia beneficiaries in hospice in 2018. CONCLUSIONS AND IMPLICATIONS: Collectively, International Classification of Diseases, Tenth Revision (ICD-10) coding and sequencing conventions, coding practices, and research methods related to hospice claim diagnoses may unintentionally underestimate and oversimplify how dementia manifests in hospice utilization.

Disaggregating the dementia monolith: An analysis of variation in Medicare costs and use by dementia subtype.

IMPORTANCE: With an aging population, it is necessary to systematically examine variation in costs and use of Medicare services by dementia subtype. We provide the first national estimates for dementia by subtype, and the respective Medicare costs and use. METHODS: We analyzed Medicare fee-for-service (FFS) claims from 2017 through 2019. The sample included 41 million beneficiaries: 727,700 beneficiaries with a new dementia diagnosis in 2017. We calculated descriptive statistics and conducted generalized linear regression models by subtype of dementia. RESULTS: Annual Medicare costs for beneficiaries with dementia ranged from $22,840 for frontotemporal dementia to $44,896 for vascular dementia compared to $9,034 for beneficiaries without dementia. Comparing beneficiaries across dementia subtypes, the greatest differences were in the use of home health and hospice care. CONCLUSIONS: These analyses demonstrate substantial heterogeneity across dementia subtypes, which will be important in developing models of care that improve value for people with dementia.