Factors Associated With Poor Outcomes in Patients who Receive Cancer Directed Therapy at the End of Life

An abstract published in 2017 that investigated more than 8000 patients across 10 tumor types who were treated at two large academic centers in the United States between 2014-2016 reported that between 3% and 7% of cancer patients received chemotherapy within 14 days of death and between 6% and 16% received chemotherapy within 30 days.¹ While chemotherapy use in cancer care is crucial for curative, noncurative, and palliative purposes, the appropriate use of cancer-directed therapy (CDT) near end of life (EoL) remains unclear.² Patient performance status (PS) often is used as a surrogate marker of anticipated tolerance and response to chemotherapy and, therefore, used to identify whether chemotherapy will offer clinical value.³ Two studies published in the 1980s that were aimed at evaluating prognostic factors in cancer care identified that chemotherapy use in patients with poor PS yielded decreased survival and response rates and high toxicity rates; however, these studies did not include some of the less toxic options approved in recent years, such as immunotherapy and monoclonal antibodies.^4-5 Recently in 2012, the American Society of Clinical Oncology published the Top Five List for Oncology, which recommends against the use of chemotherapy in solid tumor patients who have an Eastern Cooperative Oncology Group (ECOG) PS score ≥3 and have not benefited from prior evidence-based CDTs. Despite the rapidly changing landscape of CDT, there are limited options with robust evidence of benefit in later lines of therapy for most advanced cancers. However, all interventions pose the potential risk for toxicity and reduction in quality of life (QoL).⁶

Patient prognosis remains complex with limited available prognostic tools.² Therefore, the decision to start or stop treatment in favor of best supportive care remains challenging, and inappropriate antitumor interventions in refractory and terminal patients are commonly used.² Prevalence and appreciation of palliative care services has improved understanding about available alternatives to CDT in patients with advanced disease, but current published trends suggest that “patients are increasingly receiving chemotherapy within the last 2 weeks of life, requiring more frequent visits to the hospital and emergency department in the last month of life, and are often referred to hospice care in the last days of life as a means of managing death, rather than as a tool for palliation of symptoms in the later months of advanced disease.”⁶ Discussions aimed at setting goals of care, prognosis, realistic outcomes following available CDT, and available alternatives to CDT in the setting of a terminal illness happen late in the course of therapy or not at all.⁶ Given the rising costs of cancer care, this aggressive approach raises concerns regarding health care system sustainability.² Overall, limited guidelines are available to assist with the management of cancer patients nearing EoL.

Despite requiring higher acuity of care near the EoL, a paucity of literature exists determining the patient, hospital, and treatment factors most predictive of mortality amongst patients receiving chemotherapy near EoL. Accordingly, the purpose of this study is to identify factors that are predictive of early death after chemotherapy administration within the last weeks of life. Identification of these factors is thought to then assist in the creation of standardized assessments to guide the clinical decision-making on whether to administer EoL CDT vs palliative or comfort care.

Selection and Description of Participants: A retrospective review of adult inpatients who received CDT between January 1, 2015 and June 30, 2019 was conducted and approved through the investigational review board. Patients who met the following criteria were included in the study: 1) aged ≥18 years; 2) received intravenous (IV) chemotherapy, monoclonal antibodies, or immunotherapy in the inpatient setting; and 3) died ≤30 days after administration of CDT. Exclusion criteria included: 1) patients who received chemotherapy as part of a conditioning regimen for stem cell transplant or chimeric antigen receptor (CAR) T cell therapy; 2) enrolled in a clinical trial; 3) prisoners; or 4) pregnant women.

Variables: The following demographic data were collected: age, gender, race, and marital status. Vital signs and laboratory values at the time of CDT administration were analyzed and included: temperature, blood pressure, heart rate, respiratory rate, albumin, serum creatinine, alanine transaminase, aspartate transaminase, complete blood count with differential, and microbiology cultures. The presence of hepatic or renal dysfunction was graded according to the Clinical Toxicity Criteria of Adverse Events (CTCAE) v.5.⁷ Other collected data included: cancer diagnosis and stage, number of previous lines of CDT, ECOG performance status score or Karnofsky performance status score if ECOG was not assessed, level of care at the time of CDT administration, code status, presence of an advanced directive, presence of a palliative care consultation within 30 days of CDT initiation, and length of stay (LOS).

Statistics: Descriptive statistical analysis was used; an adjusted multinomial logistic regression model was performed to identify factors that may be predictive of earlier death. The adjusted multinomial logistic regression model excluded ECOG scores due to inconsistencies in reporting and utilized death within 15 to30 days as the reference group. This model was performed by authors A.W. and M.W. using SAS 9.4. Reported P values correspond to a Wald test of whether there was no significant difference in the odds of dying. 

At the time of study initiation, 149 patients received 160 CDTs and met inclusion criteria. The majority of patients were male (68%), White (61%), had a diagnosis of a hematologic malignancy (59%), and were aged a median 61 years (Table 1). About 82% of patients had a recorded ECOG score at the time of CDT initiation, and the majority were assessed to have an ECOG of >3 (44%). The majority of patients did not have a planned admission for CDT (82%), and most patients received CDT on the intermediate medical unit or medical surgical unit (86%). Notably, the majority of patients seemed fit for treatment having had less than one previous line of treatment, negative microbiology cultures, and normal organ function; however, the majority of patients could also be considered unfit as the majority was tachypneic, tachycardic, thrombocytopenic, and had a mean albumin of 2.7 ±0.6 g/dL. The median LOS was 18 days with a median time to death after CDT of 13 days. 

Most patients lacked the presence of an advance directive (66%) and were full code at the time of CDT administration (74%). Only one third of patients received consultation with the palliative care service (32%) within 30 days of CDT initiation. The median time to transition to do not resuscitate status with full care or to comfort care before the date of death was 5 and 1 days, respectively. 

An adjusted multinomial logistic regression demonstrated that an albumin of <3 g/dL had a greater than five-fold increase in the risk of death in patients who died within 1 to 7 days of receiving CDT compared to our reference group, patients who died within 15 to 30 days (odds ration [OR] 5.59; 95% CI, 1.51-20.63; P=.01). Similarly, the presence of hepatic dysfunction at the time of CDT resulted in an eight-fold increase in the risk of death for patients in the 1 to 7 days group when compared with those in the 15 to 30 days group (OR 8.22, 95% CI, 2.17-31.17; P=.01).  

When comparing patients that received CDT within 8 to 14 days with those who received CDT within 15 to 30 days of death, there was a trend toward an increased risk of death in patients who were tachypneic and had hepatic dysfunction, although this was not statistically significant (Table 2). Of note, an albumin <3 g/dL and hepatic dysfunction did not result in a statistically significant increase in the risk of death when comparing patients who received CDT within 8 to 14 days compared with 15 to 30 days of death.  

While all patients in our cohort received CDT within 30 days of death, the median was within 13 days of death. Previous studies have described the use of PS as one of the key patient factors to consider when deciding to administer CDT toward the EoL.^2,3,8,9 In our study, 44% of CDT regimens (70/160) were administered in patients who had an ECOG >3; 18% of CDT regimens were administered in patients who had an undocumented PS, limiting the ability to analyze this variable. The use of CDT in patients with poor PS is generally not recommended,^9-11 though we found the majority of our patients fit into this category demonstrating that sicker patients received CDT. Our findings are in agreement with several studies that have shown that patients with poor PS receive chemotherapy, and some patients continued to receive aggressive therapy at EoL.^3,11-14 Furthermore, ECOG scores were not available for nearly a fifth of patients who received chemotherapy during an inpatient admission, which was previously described at our institution.¹⁵ Currently, this documentation is not a required component of oncology progress notes at our institution though recommended by guidelines for safe use of chemotherapy.¹⁶ This could be a larger issue as our analysis only included patients who received chemotherapy within 30 days of death and not all oncology patients. A concerted effort must be made to assess and document PS; it is especially important as an admission to the emergency department or hospital may represent a change in PS. 

While baseline organ dysfunction was infrequent, in the patients with organ dysfunction and specifically in those with elevated hepatic enzymes, the risk of death after CDT was greatly increased. This could suggest that patients with organ dysfunction were not as fit to tolerate CDT and were more likely to experience CDT-related toxicities. 

Importantly, hypoalbuminemia and abnormal vital signs were prevalent. Low albumin (<3.5 g/dL) is a recognized marker for poor prognosis in oncology patients, suggestive of malnutrition and overall poor health.^8,9 Currently, neither albumin nor vital signs are treatment parameters within our institution’s inpatient treatment plans. Additionally, treatment parameters such as platelets, creatinine clearance, and liver function tests often were adjusted or removed completely to allow for treatment in patients who did not meet those parameters. One strategy to assess inappropriate CDT would be to employ a “question order” within our treatment plans requiring providers to document the rationale for deviation from treatment parameters. 

Another observation was that the majority of patients did not have an advanced health care directive in their medical record. This could reflect the lack of patient understanding about their disease and prognosis, which may affect their decision-making regarding therapy. Standard practice should include ensuring that oncology patients have advance directives in place and have discussed their goals of care with family and caregivers. One way to accomplish this is to involve a palliative care team early after diagnosis and at progression. Concurrent oncology and palliative care is a well-established intervention that improves outcomes, such as QoL, and aids patients in making informed treatment decisions by facilitating understanding of their disease prognosis.¹⁷ When patients are making decisions about pursuing CDT, it should be after a thorough discussion with the treating physician including the realistic potential benefits and risks. Our health system is in the process of requiring documented informed consent for chemotherapy given in the hospital. Previously, verbal consent was accepted, the details of which were typically noted in the oncologist’s progress note. 

Overall, we found that several patient factors may have contributed to death after administration of CDT given in the hospital. Going forward we may suggest including vital signs and albumin as parameters for treatment and that PS is assessed and documented before CDT is administered. Our study had several limitations including the retrospective nature of our evaluation, patient QOL scores were not assessed, and details of prior lines of therapy were sometimes missing. In addition, we did not assess chemotherapy dose adjustments for renal or hepatic dysfunction. Future studies should include evaluation of early death after CDT in both the inpatient and outpatient settings. Another avenue to explore would be examining the prescribing physician’s views regarding chemotherapy use near EoL. 

Our study found that patients received CDT a median of 13 days prior to EoL and many spent the last month of their life in the hospital. The majority of these patients had ECOG >3 and may not have been ideal candidates for CDT. In our study, there were trends of poor outcomes in patients with low albumin, tachypnea, and hepatic dysfunction. Larger prospective studies are needed to confirm our findings. Several patients in our study also lacked any documented advanced care planning. Further studies are needed to evaluate the involvement of palliative care teams for comprehensive evaluation including patients’ plans of care, disease specific factors, intent of treatment, and PS scores determining appropriate use of chemotherapy.  

Authors: Yasmine Anouty, PharmD¹; Ila M Saunders, PharmD, BCOP^1,2; Shanna Block, PharmD, BCOP¹; Katherine Medley, PharmD¹; Rabia S Atayee, PharmD, BCPS^1-3; Megan Wall, MS; Alison Wall, MS; Carolyn Mulroney, MD⁴; Janine Martino, PharmD, BCOP, APh¹

Affiliations: ¹Department of Pharmacy, University of California San Diego Health (UCSDH), San Diego, CA; ²UC San Diego Skaggs School of Pharmacy & Pharmaceutical Sciences, San Diego, CA; ³Doris A. Howell Palliative Care Teams, UCSDH, San Diego, CA; ⁴Department of Medicine, Division of Bone Marrow Transplantation, UCSDH, San Diego, CA

Address correspondence to:
Yasmine Anouty
Phone: +1 (949) 395-2005
Email: yanouty@health.ucsd.edu