Factors associated with high intensity of end-of-life care for decedents with dementia in a Medicare Advantage plan.

Abstract

There are limited data on end-of-life care (EOL) for decedents with dementia enrolled in Medicare Advantage plans. The purpose of this descriptive analysis is to compare socio-demographic and clinical characteristics and patterns of healthcare utilization for decedents who were hospitalized in the last 30 days of life versus those who were not. Data were extracted from electronic health records (EHR) of individuals aged 65+ who had at least one coded ADRD diagnosis and died in 2018-2019. Decedents with one or more hospital stays in the last 30 days of life were categorized as receiving high-intensity EOL care (n = 6299, 45%); those with no hospitalizations were classified as receiving low-intensity EOL care (n = 7571, 55%). We compared socio-demographic and clinical characteristics and healthcare utilization, including life care planning (LCP) and palliative and hospice care between the two cohorts. The high-intensity cohort was minimally younger (85±7.6 vs. 86.6±7.6) and included more men (48% vs. 41%) and people of color (47% vs. 38%) compared to the low-intensity cohort. Co-morbidities were also higher in this group (Elixhauser: 11.5±3.6 vs. 9.9±3.7) and dementia was more often first documented in the EHR very "late" [median: 586 (IQR:85.0, 1521.0) vs. 1062 (IQR:391.0, 2012.0) days prior to death]. The high-intensity group used more outpatient (primary care and specialty visits, 10.3 vs. 6.5), emergency visits (2.7±3.0 vs. 1.8±2.3), and inpatient (1.6±2.2 vs. 1.1±1.5) care in the prior year. Moreover, the first LCP conversation occurred much closer to death in the high- vs. low-intensity group [median: 35 (IQR:10.0, 139.0) vs. 105 (IQR:45.0, 234.0) days from first conversation to death]; an advance directive or POLST were less likely to be completed (71% vs. 81%) as was receipt of home palliative care or hospice (62% vs. 82%) than in the low-intensity cohort. All comparisons between the two cohorts were significant, p<.01. We found notable differences in socio-demographic and clinical characteristics, and utilization patterns in the year prior to death between decedents with dementia who were hospitalized in the last 30 days of life vs. those who were not. Further analyses will assess key, potentially modifiable drivers of high-intensity EOL care.