Associations between Medicare Expenditures and Longer Hospice Duration Vary by Clinical Condition

Monday, June 24, 2019: 8:15 AM
Tyler - Mezzanine Level (Marriott Wardman Park Hotel)

Presenter: Peiyin Hung

Co-Authors: Sylvia H. Hsu; Shiyi Wang

Discussant: Lauren Hersch Nicholas

Importance: End-of-life expenditures are a driver of increasing health care costs in the United States, with hospice use assumed to be a potential solution. Yet, prior work testing this assumption with inconsistent expenditure measures and diseases has concluded mixed results.

Objective: To examine the differential associations between hospice length of stay and Medicare expenditures at the last 90, 180, and 360 days of life among newly-diagnosed patients with COPD, CHF, dementia, and cancer.

Design: This retrospective cohort study identified decedents from Surveillance, Epidemiology, and End Results-Medicare data who were diagnosed with COPD, CHF, dementia, and cancer during 2001-2013 and died in 2002-2014. We used zero-inflated negative binomial regression models to examine the differential associations between length of hospice use and medical expenditures by newly diagnosed condition.

Participants: A total of 240,246 Medicare beneficiaries with newly-diagnosed COPD, CHF, dementia, and cancer, who had died between 13-36 months after diagnosis at age 67.5–94.9 and who had been enrolled in Medicare Parts A and B during the final 24 months of life. Non-cancer patients were identified from a random 5% sample of Medicare beneficiaries. Newly diagnosed was defined as a diagnosis in inpatient, outpatient, or physician settings without previous episodes within a year.

Exposure: Length of hospice services during the last year of life

Main Outcome: Total medical expenditures (2016 dollars) in the last year of life

Results: Cancer decedents had longer lengths of hospice use (27.4% with more than 7 days), than COPD (12.6%), CHF (12.5%), and dementia (20.4%). In adjusted analyses, longer lengths of stay in hospice beyond 7 days were associated with up to $8,352 lower expenditures for cancer decedents ($56,477 vs. $64,688; P<.001), compared to their counterparts without hospice use. In contrast, adjusted end-of-life expenditures of non-cancer patients with more than 31-day hospice enrollment were highest in each condition category. Compared to the counterparts with no hospice use, dementia patients with longer than 30-day hospice enrollments with dementia spent $17,611 more in the last year of life prior to deaths. Similarly, COPD (CHD) patients with longer than 30-day hospice enrollments cost $11,185 (9,280) more in the last year of end-of life.

Conclusions and Relevance: Longer length of hospice services for sample non-cancer decedents was associated with an average increase in Medicare expenditures in the last year of life, while decreasing expenditures for cancer decedents.