Trends in Discharge to Hospice After Critical Illness Among US Medicare Beneficiaries, 2011-2023

INTRODUCTION

In the last two decades, both clinical guidelines¹ and policy² have increasingly called for integrating palliative care for hospitalized patients at high risk of death. This shift reflects growing recognition of the importance of aligning treatment intensity with patient preferences in order to improve quality of life, reduce nonbeneficial interventions, and facilitate transitions to hospice when appropriate. As a result, the number of hospitalized patients who received palliative care has increased significantly,³ especially among those admitted to an intensive care unit (ICU).⁴

Multicenter studies have shown a positive association between inpatient palliative care delivery and discharge to hospice;^5–8 however, actual estimates of recent national trends in hospice use after critical illness remain unclear. This gap has important implications: ICU care is among the most resource-intensive in the health system, and for patients with poor prognoses, alternatives to comfort-based care often involve invasive and prolonged life support.^9–11 These dynamics are further magnified among older adults, a population in which goals-of-care decisions are common and secular shifts in intensive care decision-making have a direct, outsized influence on Medicare expenditures.^8,12–14 To better support future health system resource planning as well as inform ongoing efforts to improve end-of-life care, we sought to determine trends in hospice use after admission to an ICU across the US among Medicare beneficiaries.

METHODS

We used the 100% Medicare Provider Analysis and Review (MedPAR) and Master Beneficiary Summary Files (MBSF) to identify adults aged >65 years enrolled in fee-for-service Medicare for ≥1 year who were admitted to an ICU between 2011 and 2023. To estimate national hospice use after critical illness, the primary outcome was the annual number of discharges to hospice (inclusive of inpatient and home hospice) per 100,000 Medicare beneficiaries over 65, based on enrollment counts from the MBSF. To contextualize trends in discharges to hospice after critical illness, we also determined annual rates of in-hospital mortality, 30-day mortality, median hospital length of stay (LOS), do-not-resuscitate orders (DNR, International Classification of Diseases, Ninth Edition [ICD-9] V49.86, ICD-10 Z66), and claims for palliative care delivery (ICD-9 V66.7; ICD-10 Z51.5). Finally, because hospice use is known to vary by hospital of admission, we also performed hospital-level analyses. Specifically, to standardize against each hospital’s number of short-term decedents, we calculated each hospital’s ratio of hospice discharges to their 30-day mortality annually, and identified the national median of this ratio for each year. Standardizing by 30-day mortality, a common metric for short-term hospital mortality,^15,16 allows us to compare hospice use across hospitals with differing baseline mortality risks, and to estimate the proportion of patients at high risk of death who were transitioned to hospice before death at each hospital.

Temporal trends were evaluated using Spearman rank correlation tests. Statistical testing was two-tailed, alpha=0.05. This study met criteria for exemption by Beth Israel Deaconess Medical Center (#2022P000105) and Boston University Medical Center Institutional Review Boards (#H-41346).

RESULTS

Between 2011-2023, 10,647,283 Medicare beneficiaries (mean age 77.4 [standard deviation 8.4] years; 5,223,550 [49%] female) were admitted to an ICU (Table 1). Discharges to hospice after ICU admission increased from 388 to 572 per 100,000 Medicare beneficiaries (trend test p<0.001, Figure 1A).

Table 1.

Baseline characteristics of Medicare fee-for-service beneficiaries who were admitted to an ICU, 2011-2023.

Characteristics	Patient, No. (%) (Total = 10,647,283)
Patient characteristics
Age, years (Mean±SD)	77.4±8.4
Female	5223550 (49.1%)
Race and Ethnicity
White non-Hispanic	8987638 (85.3%)
Black non-Hispanic	988163 (9.4%)
Othera	560105 (5.3%)
Social Vulnerability Indexb (Mean±SD)	0.53±0.25
Medicaid/Medicare dual eligibility	1853116 (17.4%)
Surgical patient	4105208 (38.6%)
Received invasive mechanical ventilation	2223420 (20.9%)
Hospital characteristics
Hospital size
Small (1 - 200)	2741076 (26.2%)
Medium (201 - 400)	3619895 (34.6%)
Large (> 400)	4110309 (39.3%)
Ownership
For Profit	1676295 (16.0%)
Private Nonprofit	7622837 (72.8%)
Public	1172148 (11.2%)
Teaching hospital	7416574 (70.8%)
Rural (vs. Urban)	179283 (1.7%)
Safety net hospital	2645322 (25.2%)
Hospital region
Northeast	1832500 (17.6%)
Midwest	1668947 (16.0%)
South	5278494 (50.6%)
West	1658787 (15.9%)

^a“Other” includes: Asian, Hispanic, North American Native, and Other, as per Medicare Master Beneficiary Summary File, which derives self-reported race and ethnicity from the Social Security Administration.

^bSocial vulnerability index: each patient’s census tract is based on 15 social factors, including poverty, lack of vehicle access, and crowded housing. A census tract’s index is a percentile ranking and reflects the proportion of tracts in the country that are equal to or lower in terms of social vulnerability.

Figure 1. Trends in outcomes of Medicare beneficiaries admitted to an ICU, 2011-2023.

A. The annual number of hospice discharges per 100,000 Medicare beneficiaries over 65 years of age.

B. The proportion of ICU patients who died in hospital, were discharged to hospice, died within 30 days, received a DNR order, and received palliative care.

C. Median hospital length of stay over time.

D. Among ICU patients, the annual national median in hospital rates of hospice discharge standardized to 30-day mortality (i.e., for each hospital, annual hospice discharge divided by annual 30-day mortality rate; national median is then identified).

IQR = interquartile range.

Among ICU patients, rates of discharge to hospice rates increased from 5.6% to 6.8% (p<.001), while in-hospital mortality and 30-day mortality trends remained stable (16.0 to 15.0%, p=0.96; 23.9% to 22.0%, p=0.96, respectively). The frequency of DNR orders and palliative care delivery both increased (10.6% to 24.9%, p<0.001, and 6.8% to 14.8%, p<0.001, respectively, Figure 1B), whereas median LOS was stable (6 [interquartile range, IQR: 3,10] to 5 [IQR: 3, 9] days, p=0.87, Figure 1C).

At the hospital level, the rate of hospice discharge, standardized to 30-day mortality, increased from a median of 0.19 (IQR 0.09,0.30) to 0.27 (IQR 0.15,0.41) from 2011-2023 (p=0.01, Figure 1D), indicating that a growing share of short-term decedents were transitioned to hospice prior to death. Visual inspection of all trend lines reveal that the COVID-19 pandemic (2020-2022) temporarily led to increases in in-hospital and 30-day mortality, DNR orders, palliative care delivery, and median LOS without notable increases in discharge to hospice; in 2023, all rates appeared to return to pre-COVID-19 trend trajectories.

DISCUSSION

Between 2011 and 2023, discharges to hospice after ICU admission increased across the United States, even as overall short-term mortality remained unchanged. These findings reveal rapidly evolving patterns of end-of-life healthcare utilization among older adults in the US, which may inform policymakers’ and health systems’ future infrastructure planning. Further, given the rising use of hospice after critical illness, our findings also highlight the importance of examining outcomes after hospice to guide future efforts to improve end-of-life care.

Importantly, the stable 30-day mortality rate suggests that increased hospice discharge does not simply reflect higher mortality risk among hospitalized patients but may instead represent more intentional alignment of care with patient goals. Our results extend prior work demonstrating secular changes to high-intensity care. While use of intensive care had increased among older adults from 2000-2010,^9,17,18 there is evidence that trends in intensive care have plateaued or reversed since then. For example, among Medicare beneficiaries, intensive care unit use has since stabilized, the proportion of deaths occurring in an acute care hospital decreased,¹⁷ and the use of prolonged life support (e.g., tracheostomy) has declined as well.⁷ Medicare payment reform has also led to closure of long-term acute care hospitals, leading to decreases in long-term acute care hospital use and an increase in DNR orders among patients on prolonged mechanical ventilation.⁸ Taken together, these findings illustrate measurable shifts in preferences for intensity of care among patients at high risk of death over the last decade, possibly attributable to a mix of policy/payment changes, societal attitudes, and increased focus on patient goals.

Our study has limitations. Claims-based analyses are subject to misclassification; for example, DNR orders and palliative care delivery using ICD codes have limited sensitivity but higher specificity.^19,20 However, trends analysis of outcomes such as death, hospice enrollment, and ICD codes with high specificity can still be informative. Second, our analysis was limited to beneficiaries enrolled in fee-for-service Medicare. However, this population continues to represent a substantial proportion of older adults hospitalized in the United States and provides a stable, longitudinal dataset that allows consistent evaluation of national trends over time. Finally, our analysis only captures formal hospice enrollment; the number of patients who transition to comfort-focused care without enrolling in hospice is likely higher. Hospice use is known to vary by hospital of admission,²¹ and further work is needed to better characterize the factors (patient-, hospital-, and systems-level) and outcomes associated with hospice enrollment versus comfort-focused care without hospice.

In summary, hospice discharges following ICU admission among Medicare beneficiaries increased substantially from 2011 to 2023, despite stable short-term mortality. Continued efforts are needed to evaluate the quality and equity of these transitions, ensuring that all patients nearing the end of life receive care consistent with their values, priorities, and comfort.