Palliative Care Consultation for Hospitalized Patients with Primary and Secondary Brain Tumors at a Single Academic Center

Introduction

Primary brain tumors account for 1% of new cancer diagnoses in the United States, with age-adjusted incidence of 22.6 per 100,000. ¹ Secondary brain tumors (metastases) are more common: 200,000 are diagnosed annually, affecting up to 30% of adult patients with cancer. ² Patients with brain tumors across the spectrum of pathology, treatment, and prognosis may have common palliative care needs: burdensome physical symptoms such as seizures, headache and weakness, ³ psychiatric and behavioral symptoms, ⁴ functional and cognitive decline contributing to psychosocial and caregiver needs,^5,6 and existential distress.^7,8 For many, prognosis is poor and advance care planning and end-of-life care are important priorities.^9,10 Early palliative care is recommended for patients with high-grade glioma (the most common type of primary malignant brain tumor) by the European Association of Neuro-Oncology ¹¹ and for all advanced cancer patients by the American Society of Clinical Oncology. ¹² Despite these guidelines, patients with brain tumors are infrequently referred to palliative care and are referred to hospice close to death.^13–16 Furthermore, no guidelines exist for identifying patients with brain tumors who would most benefit from palliative care.

Since 2013, all patients with solid tumors who were admitted to Mount Sinai Hospital (MSH) oncology or to hospital medicine with oncology consulting were screened for palliative care needs. ¹⁷ Patients received palliative care consultation if they met at least one of the following trigger criteria: stage IV solid tumor or stage III lung or pancreatic cancer; prior hospitalization within 30 days; hospitalization over 7 days; or active symptoms. This trigger program was associated with increased palliative care utilization, decreased 30-day readmission and chemotherapy after discharge, and increased hospice referrals. ¹⁷ In this brief report, we compared sociodemographic and clinical characteristics of three groups of hospitalized patients with primary or secondary brain tumors: (1) those who received palliative care through primary team referral, (2) those who received palliative care through oncology trigger, and (3) those who did not receive palliative care. The purpose of this study is to generate hypotheses surrounding palliative care trigger programs for neuro-oncology.

Methods

Using administrative hospital data, we conducted a retrospective cohort study of a sample of patients with a primary or secondary brain tumor. We included patients with an index MSH admission between October 1, 2012, and March 31, 2018, with an ICD-9 or ICD-10 code corresponding to any of the following intracranial tumor categories: benign, primary malignant, or secondary malignant (metastatic). By merging our administrative neurosurgical and clinical palliative care databases, we created a database of sociodemographic, clinical, and health care utilization data for the cohort. Exposure to inpatient palliative care consultation was categorized by mechanism of referral: “nontrigger” if referred through physician request, “trigger” if the primary reason for consult was “trigger,” and “usual care” for those who did not receive palliative care. Using Kruskal–Wallis or chi-square tests, we compared the three groups with respect to age, gender, race, medical comorbidity (measured by the Elixhauser Comorbidity Index), ¹⁸ tumor type, discharge disposition, and proportion of do-not-resuscitate (DNR) orders at discharge. The palliative care groups (nontrigger and trigger) were further compared with respect to days from admission to the palliative care consult, days from consult to death (for patients who died), and proportion transferred to the palliative care unit (PCU), which is most often done for patients with a prognosis on the order of days. The Icahn School of Medicine at Mount Sinai's Institutional Review Board approved the study.

Results

Of 1669 patients who met inclusion criteria, 386 (23.1%) received palliative care [“nontrigger”: 246 (14.7%); “trigger”: 140 (8.4%)] and 1283 (76.9%) received “usual care” (Table 1). The nontrigger group was older (mean age 65.0 ± 13.6 years) than both the trigger group (61.1 ± 12.6 years) and the usual care group (55.5 ± 16.8 years; p < 0.001). The groups differed significantly with respect to gender [nontrigger: 116 (47.2%) men, trigger: 71 (50.7%), and usual care: 526 (41.0%); p = 0.028] and proportion of black patients [nontrigger: 41 (17.5%), trigger: 25 (18.4%), and usual care: 73 (5.6%); p < 0.001]. Comorbidities (by Elixhauser Comorbidity index) were highest in the nontrigger group (3.76 ± 2.01) compared with the trigger group (3.49 ± 2.01) and the usual care group (1.84 ± 1.59; p < 0.001). The majority of patients who received palliative care had a secondary brain tumor (metastasis): nontrigger: 158 (66.4%), trigger: 114 (82.0%), followed by a primary malignant brain tumor: [nontrigger: 63 (26.5%) trigger: 23 (16.6%), p < 0.001].

When comparing severity of illness between groups, the in-hospital mortality rate was highest in the nontrigger group [34 (13.8%)], followed by the trigger group [10 (7.1%)] and lowest in the usual care group [7 (0.6%); p < 0.0001]. This trend was present for the percentage discharged to hospice [nontrigger: 54 (22.0%), trigger: 17 (12.9%), usual care: 14 (1.1%)]. At discharge, 148 (60.2%) patients in the nontrigger group had a DNR order, compared with 62 (44.3%) in the trigger group and 49 (3.9%) in the usual care group (p < 0.0001).

In the nontrigger group, reasons for palliative care referral were goals-of-care discussion [114 (61%)], pain management [42 (22.5%)], support/emotional distress [12 (6.4%)], nonpain symptoms [10 (5.4%)], and transfer to the PCU [2 (1.1%)]. In the nontrigger group, 69 (33.8%) patients were transferred to the PCU, compared with 22 (15.7%) in the trigger group (p < 0.001). Among patients who died in the hospital, the nontrigger group had fewer days from palliative care consult to death than the trigger group (nontrigger: median 6.0 days, interquartile range, IQR [4.0, 12.0]; nontrigger: median 13.0 days, IQR [4.0, 12.0]; p = 0.002).

Discussion

In this cohort of hospitalized patients with primary or secondary brain tumors, patients receiving inpatient palliative care consultation through primary team referral were oldest, had the most comorbidities, and had the most severe illness (highest rates of in-hospital death, discharge to hospice, DNR order on discharge, transfer to the PCU, and fewest days from palliative care consult to death), all of which implies proximity to death at the time of consultation. This is consistent with previous studies indicating that palliative care referral is infrequent and delayed in patients with brain tumors. ¹³ In our cohort, however, triggered palliative care consultation was associated with earlier referral in the index hospitalization, and consultation occurred for patients with less severe illness, suggestive of an earlier point in the disease course. The usual care group was the youngest, had the fewest comorbidities, the highest proportion of benign tumors, and the fewest markers of severe illness. These findings suggest that oncology palliative care triggers may promote earlier palliative care for patients with brain tumors.

Benefits of earlier inpatient palliative care consultation include increased patient and caregiver satisfaction with care, ¹⁹ reduced nonbeneficial care at the end of life, ²⁰ longer hospice length of stay, ²⁰ and decreased health care costs. ²¹ Differences between early and late consultation have not previously been explored for brain tumors. Yet it may have even greater importance for these patients, who commonly experience early loss of decision-making capacity.^4,22–25 Upstream palliative care referrals may prevent patients from receiving palliative care shortly before dying in the hospital or being discharged to hospice close to death, when they may not receive its benefits.^19–21

This study's limitations include its retrospective design and restriction to a single site with a well-established palliative care program. Our general oncology triggers may not capture unmet palliative care needs of patients with brain tumors and are not applied to patients admitted to services such as neurology or neurosurgery. In addition, patients with primary tumors may differ from patients with brain metastases with respect to palliative care needs, so the inclusion of metastases (secondary brain tumor) as a trigger criterion further limits generalizability.

The significant racial differences between groups in our cohort are not fully explained by brain tumor epidemiology. ¹ Another study has suggested that inpatient palliative care consultation may neutralize some racial/ethnic differences in access to palliative care and hospice that exist in the outpatient setting. ²⁶ Further research is needed to understand differences in receipt of palliative care in this population.

Our findings provide preliminary evidence that trigger programs may be associated with increased and earlier palliative care referral for patients with brain tumors, as seen in broader oncology populations.^27–29 Trigger programs in other neurologic subspecialties such as neurocritical care and behavioral neurology (dementia) have been associated with increased specialty palliative care utilization, ³⁰ greater number of palliative care domains addressed, and increased discussion of prognosis and goals of care. ³¹ Yet, brain tumor-specific trigger criteria and other strategies to improve palliative care referral have not been tested.^13,32 Symptom-based triggers may be less appropriate for this population than for patients with systemic cancers, given the predominance of neurologic symptoms that may be outside the scope of many palliative medicine specialists' training. Other palliative care domains such as spiritual/existential distress, ⁸ caregiver needs,^7,33,34 and advance care planning^4,22,35 are particularly important and could be considered when developing brain tumor-specific criteria.^36–38

Conclusion

Ultimately, little is known about the optimal model of palliative care delivery in neuro-oncology. Our findings suggest that trigger programs—tailored to institutional needs and resources—merit consideration, likely in conjunction with other measures such as increased palliative care education for neuro-oncologists. Neuro-oncology clinicians have limited palliative care training opportunities, which may contribute to lack of familiarity with its benefits and low rates of referral to palliative care for their patients.^38–40 Finally, elucidating the perspectives of patients with brain tumors, their caregivers, and neuro-oncology providers on barriers and facilitators of palliative care referral is a key step to addressing complex unmet palliative care needs in neuro-oncology that may be distinct from those of a general solid oncology population.