Palliative and end-of-life care in glioblastoma: defining and measuring opportunities to improve care

Abstract

Background

American Society for Clinical Oncology (ASCO) quality measures for terminal cancers recommend early advance care planning and hospice at the end of life. We sought to evaluate adherence to 5 palliative care quality measures and explore associations with patient outcomes in glioblastoma.

Methods

This is a retrospective analysis of 117 deceased glioblastoma patients over 5 years. Records were reviewed to describe adherence to palliative care quality measures and patient outcomes. Data regarding emotional assessments, advance directives, palliative care consultation, chemotherapy administration, hospice, location of death, and overall survival were collected.

Results

Median overall survival was 12.9 months. By the second oncology visit, 22.2% (26/117) had an emotional assessment completed. Advance directives were documented for 52.1% (61/117) by the third neuro-oncology visit (30/61 health care proxy), yet 26.5% (31/117) did not have any advance directive before the last month of life. With regard to other ASCO quality measures, 36.8% (43/117) had a palliative care consult; 94.0% (110/117) did not receive chemotherapy in the last 14 days of life; 59.8% (70/117) enrolled in hospice >7 days before death; and 56.4% (66/117) died in a home setting. Patients who enrolled in hospice >7 days before death were 3.56 times more likely to die in a home setting than patients enrolled <7 days before death or with no hospice enrollment (P = .002, [OR 3.56; 95% CI, 1.57–8.04]).

Conclusions

Late advance directive documentation, minimal early palliative care involvement, and the association of early hospice enrollment with death in a home setting underscore the need to improve care and better define palliative care quality measures in glioblastoma.

Glioblastoma is the most common type of primary malignant brain tumor with 13,000 new diagnoses in the United States each year.¹ The standard treatment for glioblastoma is maximal safe surgical resection, external beam radiotherapy, and temozolomide, an alkylating chemotherapy agent.² Although the FDA has approved several new therapies over the past decade, none have had a significant impact on longevity, and median survival remains approximately 12 to 15 months.^3,4

Glioblastoma is complicated by significant morbidity that is burdensome for both patients and caregivers.^5,6 Physical symptoms include headache, nausea, vomiting, and excessive fatigue as well as specific neurologic deficits such as seizures, motor weakness, aphasia, and impaired vision. In contrast to other malignancies, early cognitive decline is a common symptom in patients with glioblastoma.^7–9 Due to the poor prognosis, limited treatment options, high symptom burden, and early cognitive decline, medical care in glioblastoma must balance anticancer therapies with preservation of quality of life and appropriate advance care planning.

In patients with malignancies, early and systematic advance care planning is associated with less aggressive care at the end of life, improved patient and caregiver quality of life, earlier hospice enrollment, less psychologic morbidity for patients and caregivers, alignment of care with patient preferences, more cost-effective care, and improved survival.^10–14 The research surrounding these topics in glioblastoma is limited. Investigators have defined the care needs of patients and families with high-grade gliomas, but few have examined quality of life and patient outcomes.^5,6 Most of the current data includes the malignant brain tumor population at large, making the results difficult to interpret because of the prognostic variation between different tumors.

There is an increasing imperative that patients with cancer and their families receive high-quality palliative care throughout the course of their illness. In 2012, the American Society for Clinical Oncology (ASCO) implemented a practice-based quality assessment and improvement program called the Quality Oncology Practice Initiative (QOPI). This program compiled over 180 quality measures for oncology practices that focus on improving the quality of care.¹⁵ The guidelines are standardized for all oncology patients but their utility in glioblastoma is unknown. In this study, we evaluated the performance of 5 palliative care quality measures in patients with glioblastoma at a single institution. We also performed an exploratory analysis to determine if there is an association between individual quality measures and patient outcomes.

Materials and Methods

Study Setting

The University of Rochester Division of Neuro-Oncology cares for patients in a multidisciplinary outpatient clinic that includes specialists from neuro-oncology, radiation oncology, and neurosurgery. At least 30 to 40 patients with newly diagnosed glioblastoma are seen in the outpatient setting each year, and the majority are followed from diagnosis until death.

Defining the Study Cohort

We conducted an Institutional Review Board-approved, retrospective analysis of all glioblastoma patients at the University of Rochester Medical Center whose date of pathologic diagnosis was between January 1, 2010 and May 1, 2015 and whose date of death was before November 1, 2015 (n = 173). To ensure a well-defined cohort with available data, we excluded patients who received their primary oncologic care at an outside institution (n = 16) and those who were alive at the time of analysis (n = 40).

ASCO Quality Measures

We reviewed all ASCO QOPI measures related to palliative and end-of-life care. Five measures were selected based on their relevancy to the glioblastoma population and comprehensive representation of end of life care. The measures include: (1) emotional assessment by the second oncology visit, (2) advance directive documentation by the third oncology visit, (3) no chemotherapy administration within the last 14 days of life, (4) hospice enrollment >7 days before death, and (5) hospice and/or palliative care involvement at the end of life. In general, ASCO recommends at least 75% adherence to quality measures for high-quality cancer care.

Data Collection and Assessment of Adherence to Quality Measures

The primary author (LH) reviewed the relevant sections and provider notes within the electronic medical record (EMR) to obtain the following information: patient’s demographics, date of diagnosis, extent of tumor resection, pathologic diagnosis, dates of outpatient neuro-oncology visits, date of death, and quality measure adherence. Here we detail our approach to defining quality measure adherence.

1. Emotional Assessment by the Second Oncology Visit. Documentation of the patient’s emotional stress level, ability to cope with illness, or mood disturbance in either the history of present illness or assessment and plan of a neuro-oncology or palliative care note. Documentation of mood in either the review of systems or physical exam was excluded from quality measure adherence.

2. Advance Directive Documentation by the Third Oncology Visit. The presence of any of the following in the subject’s EMR: health care proxy form, living will, Medical Orders for Life Sustaining Treatment (MOLST), do-not-resuscitate (DNR) form, and/or hospice enrollment. If a study subject did not have a third oncology visit, advance directive documentation within 2 weeks of the prior visit was considered adherence to the quality measure. A MOLST form is specific to New York State and includes preferences about cardiac resuscitation, intubation, artificial feeding, intravenous fluid administration, antibiotic usage, and future hospitalization. For patients with a MOLST form completed before death, we recorded the person of consent (patient vs other).

3. No Chemotherapy Administration within the Last 14 Days of Life. Last date of chemotherapy was determined via medication administration or date of oncology visit when the decision was made to stop all chemotherapy. This could not be determined for patients receiving treatments outside of the participating institution (n = 4).

4. Hospice Enrollment >7 Days before Death. Hospice is a program available to patients with a prognosis of <6 months who want to focus on symptomatic treatment without the use of disease-modifying therapy. Patients who are enrolled in hospice are often cared for at home or in a hospice-associated facility. Individual dates of hospice enrollment were verified via scanned images of hospice enrollment papers, an encounter with hospice staff, or telephone encounters confirming hospice enrollment.

5. Hospice and/or Palliative Care Involvement at the End of Life. Hospice involvement was confirmed using the methods outlined above. Palliative care is a specialty that focuses on symptomatic and supportive care of the patient at any time in disease course. Date and setting (inpatient vs outpatient) of the first palliative care consultation were recorded.

Patterns and Outcomes of Care

Overall survival was calculated from the date of pathologic diagnosis to the date of death. Location of death was confirmed through the EMR and/or patients’ obituaries and recorded as follows: home, comfort care home (a 2-bed home for patients who have a prognosis of less than 3 months, are enrolled in hospice, and cannot stay at home for the end of life), skilled nursing facility with hospice, inpatient hospice unit (hospital unit for patients who are within their final weeks of life and require more intensive end-of-life care such as intravenous medications), skilled nursing facility without hospice, intensive care unit, and emergency department. The location-of-death categories were grouped as death in a home setting (home or comfort care home) or death in a health care facility (skilled nursing facility with hospice, inpatient hospice, skilled nursing facility, intensive care unit, or emergency department). We also further defined the timing of hospice enrollment.

Palliative Care Opportunity Analysis

We performed a palliative care opportunity analysis by looking at the proportion of a patient’s life after diagnosis spent without advance directives or treatment-limiting orders. To perform this analysis, we assessed the timing of advance directive and treatment-limiting order documentation (MOLST, DNR, or hospice enrollment) and calculated the time point that such directives and orders were put in place compared to the patients overall survival from diagnosis to death. We then plotted and analyzed the overall proportion of survival lived without advance directive and treatment-limiting orders.

Data Analysis

We used descriptive statistics to summarize the clinical characteristics of the cohort and to determine the proportion of patients adhering to each quality measure.

Kaplan–Meier survival analysis was used to determine overall survival. We used chi-square tests, Kaplan–Meier curves, and multivariate logistic and linear regressions to determine if there were associations between clinical characteristics, adherence to quality measures, and patient outcomes.

All reported P values are 2-sided, and P values <.05 are considered statistically significant.

Results

Patient Characteristics

The final cohort included 117 patients who were diagnosed with glioblastoma between January 1, 2010 and May 1, 2015, and who died before November 1, 2015. Table 1 shows the demographic and clinical characteristics of the study patients (n = 117). The median age was 63 years; 57.3% (67/117) were male; and 37.6% (44/117) had a gross total resection of their tumor. Thirty-seven percent (43/117) of patients had a palliative care consult during their care, the majority of which took place during an inpatient hospital admission (72.1%, 31/43). The median survival was 12.9 months (range, 1.0–42.8 months). Greater overall survival was associated with gross total resection of tumor (P<.0025) and age <60 years at diagnosis (P<.0001).

Table 1.

Patient Characteristics (N = 117)

Variable	Median (Range)	Total Patients (%)
Age, years	63 (27–84)
Male Gender		67 (57.3)
Surgical Intervention
Biopsy Only		30 (25.6)
Partial Resection		43 (36.8)
Gross Total Resection		44 (37.6)
Inpatient PC Consult		31 (26.5)
Outpatient PC Consult		12 (10.3)
Median Survival (Months)		12.9

Abbreviation: PC, palliative care.

ASCO Quality Measure Adherence

Table 2 depicts the total number of patients meeting each ASCO quality measure (n = 117). Adherence ranged from a low of 22.2% (26/117) for the emotional assessment measure to a high of 94.0% (110/117) for the no-chemotherapy-in-the-last-14-days measure. The majority of the emotional assessment documentation was done by the neuro-oncology group (24/26) compared with the palliative care consultants (2/26). Of the 3 patients who received chemotherapy in their last 14 days of life (3/117), 2 received bevacizumab and 1 received a non-chemotherapeutic agent on a clinical trial that was thought to have antitumor effects. Chemotherapy administration was unknown for 3.4% (4/117) of patients.

Table 2.

ASCO Quality Measure Adherence (N = 117)

Quality Measure	Number of Patients Meeting Quality Measure (%)
Emotional assessment by second oncology visit	26 (22.2)
Advance directive documentation by third oncology visit	61 (52.1)
No chemotherapy in the last 14 days of life	110 (94.0)
Hospice enrollment >7 days before death	70 (59.8)
Hospice and/or palliative care at the EOL	98 (83.8)

Abbreviations: ASCO, American Society for Clinical Oncology; EOL, end of life.

By the third neuro-oncology visit, 52.1% (61/117) of patients had an advance directive documented. Most of these were health care proxy forms alone (49.2% [30/61]), followed by MOLST forms (36.1% [22/61]), living will plus health care proxy (13.1% [8/61]), and non-hospital DNR (1.6% [1/61]). Of the patients with a completed MOLST at any time before death, 45.3% (29/64) were consented for by proxies. The majority of patients died with hospice and/or palliative care involvement at the end of life (83.8% [98/117]). Sixty percent (70/117) of patients enrolled in hospice >7 days before death and the median time from hospice enrollment to death was 18.5 days (range, 0.1–7.8 months).

Patterns and Outcomes of Care

Fig. 1 shows the number of patients who died in each location (n = 117). More patients died in a home setting (56.4% [66/117]) than in a health care facility (35.9% [42/117]). There was no difference in survival for patients who died in a home setting compared with those who died in a health care facility (P= .22). Patients who died in a home setting spent more time on hospice compared with patients who died in a health care facility (median of 22 days vs 7.5 days, P = .0002)(Fig. 2, n = 90). In addition, patients who enrolled in hospice >7 days before death were 3.56 times more likely to die in a home setting than patients enrolled <7 days before death or with no hospice enrollment (P = 0.002, [OR 3.56; 95% CI, 1.57–8.04]).

Fig. 1.

Location of death categorized by home setting vs health care facility. Abbreviations: CC, comfort care; ED, emergency department; SNF, skilled nursing facility; ICU, intensive care unit; IP, inpatient.

Fig. 2.

Kaplan–Meier analysis of time from hospice enrollment to death stratified by location of death (home setting vs health care facility).

Palliative Care Opportunity Analysis

The median time from first advance directive documentation to death was 3.9 months (range, 0.1–32.2 months) and 44 of 117 patients spent more than 80% of their life after diagnosis without an advance directive (Fig. 3a, n = 117). Twenty-seven percent (31/117) of patients did not have any advance directives before their last month of life. The median time from first treatment-limiting order to death was 27 days (<1 month; range, 0.1–15.4 months) and 61/117 patients spent more than 90% of their life after diagnosis without a treatment-limiting order (Fig. 3b, n = 117). Nearly one-quarter (29/117) of patients did not have a detailed treatment-limiting order before their last week of life.

Fig. 3.

Percentage of life after diagnosis spent without an advance directive or treatment-limiting order for each individual patient. (a) Advance Directive: 44/117 patients spent more than 80% of their survival without an advance directive. (b) Treatment-Limiting Order: 62/117 patients spent more than 90% of their survival without a treatment-limiting order.

Discussion

Early advance care planning, advance directive documentation, and hospice involvement are essential to align care with patient preferences and to provide cancer patients and their families with high-quality care.^10–14 In glioblastoma and other advanced malignancies, valuable patient outcomes include minimal hospitalizations, limited interventional care, earlier hospice enrollment, and death outside of a hospital.^16,17 These principles are well-recognized in palliative care and have been recently addressed in oncology through the ASCO QOPI quality measures. In our study, we analyzed adherence to the established palliative care quality measures, explored associations with patient outcomes, examined opportunities to improve care, and reflected on the utility of those measures in glioblastoma.

We observed that over half of patients met the quality measure and had an advance directive by the third oncology visit, significantly lower than the recommended 75% adherence rate. Most of these early advance directives were health care proxy forms alone, and for many patients, this was the only indirect evidence of advance care planning prior to the end of life. Although the median time from advance directive documentation to death was 3.9 months, a substantial number of patients spent more than 80% of their survival without any advance directives. In addition, half of patients did not have a detailed treatment-limiting order prior to the last month of life and over 40% of MOLST forms were completed by proxies. Most patients with glioblastoma have impaired decision-making capacity in the last month of life.^18–22 Our data suggest that many of the advance directives in our study were completed far too late in disease course, possibly after the patient lost decision-making capacity. Palliative care quality measures in glioblastoma should emphasize early end-of-life discussions, define detailed advance directives, and focus on early directive documentation in order to protect patient autonomy and reduce the decision-making burden for caregivers.^23,24

Early palliative care has been shown to reduce symptom burden, decrease rates of depression in patients and caregivers, reduce costs of care, minimize hospitalizations and in-hospital deaths, decrease aggressive end-of-life care, and improve a patient’s survival.^{11,12,25–28} Despite these proven benefits, very few patients in our study had a palliative care consult. For those who were involved with palliative care, the majority of consults occurred during an inpatient hospital admission for an acute medical problem. Our results are consistent with the literature in illustrating that early involvement of palliative care services is rare in neuro-oncology.²⁹ In relation to our study, early palliative care interventions may be particularly valuable in improving early advance directive documentation and hospice enrollment, thus potentially improving patient outcomes. A prospective study looking at the utility and feasibility of early palliative care in glioblastoma would be helpful in determining whether palliative care interventions should be incorporated into future quality measures.

Hospice has also proven to be most beneficial to patients and caregivers when initiated earlier in the course of a patient’s illness. Over 80% of our patients had hospice or palliative care involvement at the end of life, exceeding the ASCO QOPI recommendation; however, only half of our patients enrolled in hospice >7 days before death. Earlier hospice care is associated with fewer hospitalizations and futile medical services at the end of life, better quality of life for terminally ill patients and families, and a more cost-effective approach to care.^10,16,30,31 Late hospice referral (within 7 days of death) was associated with increased use of medical services at the end of life in a cohort of primary malignant brain tumor patients.³² Despite the well-documented benefits of earlier hospice involvement, a substantial number of patients in our cohort were not receiving hospice services before their last week of life, depriving patients and families from the most comprehensive end of life care. For some cancer patients, ongoing treatments interfere with hospice enrollment, but in our cohort 94% of patients were not receiving disease-directed therapy in the last 2 weeks of life. Future prospective studies should examine potential barriers to hospice enrollment in glioblastoma, focusing on the perspective of patients, caregivers, and health care providers.

Many patients and families faced with terminal illness prefer death at home or outside of the hospital.^16,33,34 Although half of our patients died in a home setting, one-third died in a healthcare facility despite hospice involvement. Almost all of these patients were admitted to the hospital for clinical decline, subsequently enrolled in hospice, and then discharged to either skilled nursing facility hospice or inpatient hospice for the end of life. Their route to hospice was through a crisis rather than a collaborative outpatient decision to enter into hospice care, and this resulted in the unfavorable outcome of death in a health care facility. Additional studies are needed to evaluate risk factors for hospitalization at the end of life and barriers to death in a home setting for patients with glioblastoma.

Ideally, quality measure adherence should improve patient outcomes. Patients who met the quality measure and enrolled in hospice >7 days before death were almost 4 times more likely to die in a home setting. To our knowledge, these are the first data in the glioblastoma population linking earlier hospice enrollment with death in a home setting. There was no difference in survival between patients who died in a home setting and those who died in a health care facility, underscoring previous studies linking high health care utilization at the end of life with little-to-no survival benefit and likely a decreased quality of life for patients with terminal malignancies.^11,31,35 These findings may help inform future quality measures addressing the timing of hospice enrollment in the glioblastoma population and the importance of minimizing interventional care at the end of life.

The patients in our study had continuity throughout their care and were followed from diagnosis until death, allowing us to effectively study palliative care quality measures and end-of-life outcomes. The demographics and clinical characteristics of our study sample are consistent with population-based studies in the U.S.³⁶ However, the study was performed at a single institution and the practice patterns ultimately reflect the medical care provided by a handful of providers, somewhat limiting the generalizability of our results. In addition, the final study analysis may misrepresent end-of-life care in glioblastoma because patients who were alive at the time of the analysis were excluded. However, 70% of the cohort was included in the final analysis, subjects were identified by diagnosis (not death), and subjects were followed until death (not for a prespecified time period).³⁷ Finally, very few patients survive glioblastoma and a cohort of deceased patients is likely reflective of the vast majority of glioblastoma patients.

The design of our study leaves room for unmeasured confounding variables that cannot be accounted for due to the nature of a retrospective analysis. For this reason, we are unable to assess how variables like patient preferences and behaviors may have affected adherence to quality measures and patients’ outcomes. We recognize the importance of this limitation and note that these variables would be most appropriately assessed in a prospective study. Also, patients may have received some aspects of care at outside institutions that could not be captured through our EMR. Finally, we are aware that gaps in documentation may exist, including but not limited to the data about emotional assessments. To minimize the gaps in documentation and better address patient distress, we advocate for the use of standardized emotional assessment tools throughout the course of care in glioblastoma.

Our study illustrates both the complexity of assessing quality measures in glioblastoma and the need for more specific guidelines regarding advance care planning and end-of-life care in this population. Quality measures in glioblastoma should focus on defining early advance directive documentation, suggesting appropriate timing for hospice enrollment, and determining which patients may benefit from early palliative care interventions. Additional prospective studies are needed to inform these quality measures and guidelines, with the ultimate goal of improving care and outcomes for patients with glioblastoma.

Funding

Year-out research fellowship grant from the University of Rochester Office of Medical Education awarded to Lauryn E. Hemminger.

Conflict of Interest statement. Nimish A Mohile, MD, MS and Jennifer Serventi, PA-C, MS report consulting fees from Novocure outside the submitted work. There are no disclosures from other authors.