Characteristics of Hospices Providing High-Quality Care

Introduction

The Department of Health and Human Services Office of the Inspector General has highlighted concerns about variation in the quality of hospice care, and championed using data to identify hospices that need improvement or additional federal oversight. ¹

Patterns of hospice care have changed substantially over time, and vary greatly across hospice organizations. For example, the mean length of stay in hospice increased from 53.5 days in 2000 to 88.6 days in 2017; in that year, the median hospice discharged ∼1 in 5 patients alive (19.1%), but a 10th of hospices had a live discharge rate of 53% or more. ² In 2014, 8% of hospices provided no visits from professional staff during the last two days of their patients' lives, whereas a quarter of hospices provided such visits to 97% or more of their patients. ³

Prior research has shown that care patterns and populations served are notably different by hospice characteristics. For example, smaller hospices have longer lengths of stay, higher live discharge rates, and higher rates of patients receiving no general inpatient (GIP) or continuous home care (CHC). ⁴ Hospices in existence for three years or less, or located in rural areas, are less likely to provide visits from professional staff in the last days of life than older hospices and those located in urban areas. ³ For-profit hospices exhibit higher rates of live discharge,^4,5 use less-skilled staff,^6,7 provide more of their care to noncancer patients^4,6,8,9 and in nursing homes,^4,10 and have longer average lengths of stay than nonprofit hospices,^4,6,9–11 with care patterns and populations varying substantially by hospice size and chain affiliation across ownership categories. ⁴

Until recently, no systematic national data have been available to assess whether these observed variations have an impact on hospices' quality of care. This brief report uses national data from the Hospice Quality Reporting Program (HQRP) to examine factors associated with hospice care quality, assessed by measures of patient- and family-centeredness derived from the Consumer Assessment of Healthcare Providers and Systems (CAHPS^®) Hospice Survey, and measures of clinical processes from the Hospice Item Set (HIS).

Methods

Under the HQRP, hospices must submit HIS measure data and contract with a survey vendor to collect the CAHPS Hospice Survey lest they face a payment penalty. In August 2017 and February 2018, respectively, the Centers of Medicare & Medicaid Services (CMS) began publicly reporting HIS and CAHPS Hospice Survey measure scores for U.S. hospices on the Hospice Compare website. ¹²

Hospice-level HIS and CAHPS Hospice Survey measure scores were analyzed for the first period in which data were publicly reported for both. This included HIS measures from April 2016 to March 2017, and CAHPS Hospice Survey measures from April 2015 to March 2017. Details regarding HIS and CAHPS Hospice Survey measures are available elsewhere.^12–15 The average hospice response rate to the CAHPS Hospice Survey during this timeframe was 33%. Analysis was restricted to hospices that had at least one reported HIS measure and one reported CAHPS Hospice Survey measure. In keeping with public reporting minimums, HIS measures are reported for hospices with at least 20 patient stays, and CAHPS measures are reported for hospices with at least 30 survey responses. Hospices in Puerto Rico were excluded.

We examined hospice characteristics previously shown to be associated with differences in hospice care patterns.^4–11 We used each hospice's CMS Certification Number (CCN) to determine its geographic region, and a 2016 Provider of Services (POS) file and a September 2017 CMS Active Agency List to derive ownership status and hospice age; documentation submitted by hospices to CMS when authorizing a CAHPS Hospice Survey vendor to identify a hospice's chain status, with hospices determined to be a chain if they had at least two CCNs with common management; and 2015 Medicare hospice claims to calculate hospice size (number of patients cared for during the year). Hospices were defined as rural if >80% of patients in the 2015 Medicare hospice claims files lived in a rural zip code and the POS indicated that the hospice was rural. Ownership and chain were combined into a single variable to allow for assessment of the combination of ownership and chain status on care quality.

Medicare hospice claims from 2015 were used to calculate characteristics of patients under the care of the hospice (mean length of stay and percentage of patients living in a nursing home) and processes of care associated with care quality (proportion of patients discharged alive, decedents who did not receive GIP or CHC in the last seven days of life, and decedents in routine hospice care who did not receive visits by professional hospice staff in the last two days of life).^3,5,16

Summary CAHPS measure scores were calculated for each hospice by averaging each hospice's performance across CAHPS measures; in this calculation, the six composite measures assessing specific aspects of care experience received equal weight, whereas the two global assessment measures, overall rating and willingness to recommend, each received half weight, as they measure the same construct. Summary HIS measure scores were calculated for each hospice by averaging across all HIS measures, assigning equal weight to all seven measures. For hospices that were missing scores for a given measure, national mean scores were imputed for the measure. Separate logistic regression models predicted whether hospices were in the top quartile of CAHPS Hospice Survey summary measure performance, HIS summary measure performance, or both, using each of the characteristics already described as predictors.

Results

The 2746 hospices in our analyses (i.e., hospices that met the criteria for public reporting) were more likely than those excluded to be larger, older, nonprofit and not part of a chain, and located outside of the South (Table 1).

One hundred fifty-three (5.6%) of hospices were in the top quartile of performance for both CAHPS and HIS. These high-performing hospices were more likely to be medium-sized (91 to 200 patients per year; adjusted odds ratio [AOR] 2.01, confidence interval [CI] 1.36–2.96), in business for 20+ years, and serve rural areas (AOR 1.79, CI 1.20–2.68; Table 2).

Hospice characteristics associated with being in the top quartile for CAHPS included being a nonprofit and nonchain or government hospice (AOR 1.53, CI 1.11–2.12; AOR 1.96, CI 1.13–3.42), smaller size (<200 patients per year), being in the South (AOR 2.14, CI 1.57–2.93), serving a rural area (AOR 2.59, CI 2.02–3.31), and providing professional staff visits in the last two days of life to 71.1% or more patients (AOR 1.83, CI 1.40–2.38).

Characteristics associated with being in the top quartile of HIS performance included being a for-profit chain (AOR 1.95, CI 1.47–2.59), hospice age of 20+ years, larger size (91+ patients per year), having <40% of patients in a nursing home (AOR 1.40, CI 1.11–1.78), and providing professional staff visits in the last two days of life to 71.1% or more patients (AOR 1.30, CI 1.03–1.64).

Discussion

Hospice care has evolved from its roots as a nonprofit community service delivered by volunteers to a mainstream, increasingly for-profit service delivered to about half of all Medicare decedents. ¹⁶ We find that smaller independent nonprofit and government hospices are more likely to provide better patient- and family-centered care, but that large for-profit chain hospices perform better on process measures assessed through chart documentation. Our findings suggest that these dimensions of care may be largely independent, a result that is in keeping with those in other settings of care.^17,18 Notably, a small subset of hospices perform well on both types of measures, suggesting that there is no inherent tradeoff between the two and that it is possible for hospices to excel at both processes and patient- and family-centered care.

This is the first national study reporting the association between organizational characteristics and care processes and bereaved family members' perceptions of the quality of hospice care. Our results confirm that a process—visits in the last two days of life—is associated with high-quality performance, an actionable finding given high variation across hospices in the proportion of patients receiving these visits. ³

Our study does not include the smallest hospices, as quality measure data cannot be reliably calculated and, therefore, are not publicly reported, for these hospices; consequently, our results are not representative of all hospices nationwide. In addition, some have expressed concern that hospices perform so well on the HIS measures analyzed here that these measures have limited ability to distinguish performance across hospices, and that it is unclear how much the processes that HIS measures assess result in the ultimate desired outcome: high-quality care from the perspective of patients and families. ¹⁶ CMS is currently working to replace these measures with an assessment tool that examines patient outcomes. Further research should examine whether the characteristics associated with CAHPS and HIS performance also predict success on the next generation of hospice quality measures, or on other facets of care quality not assessed by national quality metrics.

As the composition of the U.S. hospice market continues to change, it is critically important to understand the organizational characteristics and clinical care processes associated with high-quality hospice care. Differences in care quality by hospice characteristics suggest opportunities for improvement, notably in the area of professional staff visits in the last days of life.