1. Introduction
The Institute of Medicine (IOM) report, “Dying in America,” detailed marked inadequacies in the care of patients at the end of life (EOL), including deficiencies in communication and advance care planning.1 Timely discussions about EOL care and documentation of patient preferences in the health record are particularly lacking for those with metastatic breast cancer (MBC).2,3 Although patients who discuss EOL care with clinicians are more likely to have documented advance directives, enroll in hospice earlier and die at home,2,4 patients with MBC often fail to receive such high quality care at the EOL.5–7 Rather, this population is at increased risk of receiving chemotherapy in the last weeks of life5,8,9 with suboptimal hospice utilization and many dying in the acute care setting.2,5,6,10,11
A key recommendation from the IOM report was for clinicians to conduct and document conversations about EOL care preferences for patients with serious illness.1 Prior research has demonstrated that the integration of palliative care (PC) from the time of diagnosis for patients with poor-prognosis cancers improves communication with clinicians and delivery of EOL care, including greater documentation of advance care plans and earlier referral to hospice before death.12–17 National organizations now recommend early PC for patients with advanced cancer and/or high symptom burden.18,19
For patients with MBC, many of whom have long disease trajectories, integrating PC from the time of diagnosis may not be feasible. Unlike the disease course of other metastatic solid tumors, MBC has demonstrated substantial improvements in survival in recent years due to treatment advances.20 Considerable variability in breast cancer prognosis also exists likely due to heterogeneity in disease subtypes, underscoring the need to identify subroups among women with MBC with poorer prognosis who may benefit from PC. Yet, defining the subpopulation of patients with MBC who may benefit from outpatient PC is challenging given that published criteria for such referrals employ general indicators including severe symptoms, poor prognosis, brain metastases, etc.21,22 To date, no prospective studies have evaluated a PC intervention specifically for patients with MBC, and trials of early PC with mixed cancer samples have included a minority of patients with breast cancer.23–27 Consequently, a gap in clinical research and practice exists with respect to the provision of PC for patients with MBC.
For this study, we identified patients with MBC who had clinical indicators of increased risk of death within six months. The primary study aim was to determine the effect of a structured PC intervention on documentation of EOL care discussions in the electronic health record (EHR). We also assessed the effect of the intervention on patient-reported discussions about EOL care preferences, QOL, and symptoms of anxiety and depression, as well as hospice utilization.
2. Methods
2.1. Study Design:
From 05/02/2016 to 12/26/2018, we enrolled 120 patients with MBC in a single-site, randomized controlled trial of a PC intervention versus usual care. The Dana-Farber/Harvard Cancer Center IRB approved this study.
2.2. Participants:
Eligible participants included patients with MBC who were within eight weeks of any of the following clinical indicators signifying poor prognosis. We drew on published criteria for referral to outpatient PC in oncology21 and solicited feedback from the Cancer Center breast oncology clinicians (physicians=9, nurse practitioners=4) to identify these indicators:
Leptomeningeal disease
Progressive brain metastasis after initial radiation therapy
Brain metastases and starting whole brain radiation
Discharged after an unplanned hospital admission
Triple negative disease and starting second-line chemotherapy
Received at least three different treatment regimens in a 12-month period
HER2+ or ER+ disease and starting third-line therapy
Beginning a treatment clinical trial
Participants were also required to have an Eastern Cooperative Oncology Group (ECOG) performance status of 0–2, receive their cancer care at the participating institution, and be able to complete questionnaires in English or with minim al assistance. Exclusion criteria included: already receiving PC or needing immediate palliative or hospice care; or serious mental illness/cognitive impairment interfering with the ability to participate, per the treating oncologist.
2.3. Study Procedures:
To identify eligible patients, the research assistant (RA) reviewed the breast oncology clinic schedules. The RA then emailed the oncology clinicians to request permission to approach their patients for study participation. Upon receiving permission, the RA approached potentially-eligible patients in clinic to explain the study procedures, obtain written informed consent, and administer baseline assessments. Patients were then randomized to the intervention or usual care group using a computer-generated randomization schema stratified by the categories of prognostic indicators per the eligibility criteria (i.e., central nervous system disease, unplanned hospitalization, or cancer treatment changes). Because patients were randomly assigned, the oncology clinicians could treat patients in both study groups.
2.4. Intervention Procedures:
Intervention:
We adapted the intervention manual from our prior research evaluating an integrated palliative and oncology care model17,28 based on feedback from two focus groups with breast oncology clinicians (physicians=9, nurse practitioners=4) and clinical observations of PC clinicians (n=2). The lead study PC clinician conducted two half-day trainings with the participating PC physicians and nurse practitioners in administering the study protocol and content of the intervention visits. The study intervention included five structured visits addressing the following: rapport building, symptom management, illness understanding, coping, treatment decision-making, and EOL/advance care planning (Appendix 1). The PC clinicians received an email prior to each structured visit reminding them of the intervention content. No clinicians received training in documenting patient EOL care preferences.
Intervention patients participated in their first PC visit within four weeks of enrollment. Subsequent intervention visits occurred every four weeks until patients completed the five structured visits (i.e., approximately 20 weeks). PC visits were scheduled on the same day as oncology visits to minimize trips to the clinic. For the first visit, we encouraged the PC and oncology clinicians to meet together with the patient. If a visit was unable to be scheduled within four weeks of the prior visit, the PC clinician contacted the patient via telephone. After each study visit, PC clinicians completed an online survey denoting the content addressed during the encounter. Once the five-visit study intervention ended, patients could request ongoing, unstructured PC visits.
Usual Care:
Patients assigned to usual care met with a PC clinician only upon request of the patient/caregiver or oncologist. When these patients received PC, they did not follow the intervention protocol.
2.5. Study Measures:
Documentation of EOL Care Preferences (primary outcome):
To identify clinician documentation of EOL care discussions, we utilized Natural Language Processing (NLP) methods to query the EHR with a validated algorithm identifying terms for goals of care and EOL discussions.29–32 The NLP software, ClinicalRegex, displays clinical notes that contain phrases about EOL discussions. An independent coder (i.e., an oncologist blind to group assignment) reviewed the documentation highlighted by NLP to ensure accurate identification of EOL care discussions. We also searched the record for completion of the Massachusetts Medical Orders for Life Sustaining Treatment (MOLST) form. The timeframe for this EHR query was from patient enrollment until date of death, last follow-up, or data cutoff for analysis among those still alive.
Participant-Reported Measures (secondary outcomes):
At baseline prior to randomization, participants completed a socio-demographic questionnaire and the following assessments. These measures were repeated at weeks 6, 12, 18 and 24 post-baseline.
EOL Care Discussions & Prognostic Awareness: We administered items used in our prior PC studies13,14 including: “Have you and your doctor discussed any particular wishes you have about the care you would want to receive if you were dying?” and “My cancer is curable” (Yes/No). For these outcomes, we analyzed the 24-week assessment or final assessment prior to patient death or transfer of care, etc., controlling for the baseline values.
QOL: We used the 37-item Functional Assessment of Cancer Therapy-Breast (FACT-B) scale,33 which assesses physical, functional, emotional, and social well-being, as well as breast cancer-specific concerns, over the past week. Higher scores indicate better QOL (range: 0–148; Cronbach’s alpha=.90).
Anxiety and Depression Symptoms: Patients completed the 14-item Hospital Anxiety and Depression Scale (HADS),34 which includes subscales for anxiety and depression symptoms in the past week, each ranging from 0 (no distress) to 21 (maximum distress). Mean internal consistency estimates are .83 for the anxiety subscale and .82 for the depression subscale.35
Health Record Review:
We manually extracted data from the EHR regarding tumor biomarkers, presence of brain metastasis, smoking status, ECOG performance status, cancer treatment, and hospice referral.
2.6. Statistical Analysis:
We performed statistical analyses on data obtained through 07/01/2019 using SPSS (v.25.0), STATA (v.14.2) and R (v.3.6.2). To summarize participant characteristics, we first calculated descriptive statistics. Following the intent-to-treat principle, we used the Fisher’s exact test and logistic and linear regression modeling to examine group differences in rates of documentation of EOL care discussions and time to documentation from enrollment, respectively. These analyses controlled for patient age given the baseline imbalance in this variable. We also conducted logistic regression to examine patient-reported discussions with clinicians about EOL care and perceptions of cure, controlling for age and baseline ratings of these outcomes. All p-values were two-sided, with p<0.05 considered statistically significant.
To examine group differences in QOL and symptoms of anxiety and depression across all assessment time points, we used the terminal-decline joint modeling approach, which accounts for missing data and models the trend in outcomes backward from death (or from date of last follow-up or data cutoff for analysis among those alive) rather than prospectively from enrollment.36 Terminal-decline joint modeling is an advantageous approach to account for deterioration in patient-reported outcomes closer to death while using a mixed-effects model for longitudinal outcomes to provide valid estimates for missing data.36 All models controlled for age and baseline scores of the outcomes. We did not adjust for multiple testing, as the secondary outcomes were exploratory and hypothesis-generating.
To evaluate differences in hospice utilization rates, we conducted the Fisher’s exact test followed by logistic regression, controlling for patient age. We also used the Mann-Whitney U test to examine days on hospice between groups.
Enrolling at least 50 patients per group provided 80% power to detect an increase from 20% to 48% in documentation of EOL care discussions. To account for missing data due to patient withdrawal and death, we increased the sample size from 100 to 120 patients.
3. Results
3.1. Participant Characteristics
Of 177 patients eligible for participation (Figure 1), we approached 92.7% (164/177) and enrolled 73.2% (120/164). Table 1 and supplemental Appendix 2 detail the sample characteristics. Of these patients, 85.8% (103/120) had previous early-stage breast cancer, and 14.2% (17/120) presented with advanced disease. At the data cutoff date, 58.7% (64/109) of patients had died, and 11 patients had unknown survival status (i.e., lost to follow-up, withdrew, or transferred care).
Figure 1.
Study Flow Diagram
Table 1.
Baseline Demographic and Clinical Characteristics
Variable | Usual Care N=59 | Intervention N=61 |
---|---|---|
Age, Mean (SD) | 58.54 (11.63) | 55.33 (10.71) |
Female | 59 (100.0) | 61 (100.0) |
Other | 2 (3.4) | 2 (3.3) |
Missing | 0 (0) | 1 (1.6) |
Missing | 1 (17) | 0 (0) |
Widowed | 7 (11.9) | 4 (6.6) |
Graduate school | 21 (35.6) | 18 (29.5) |
HER2 Positive | 12 (20.3) | 11 (18.0) |
Brain Metastasis, N (%) | 16 (27.1) | 16 (26.2) |
Unknown | 11 (18.6) | 10 (16.4) |
2 | 7 (11.9) | 4 (6.6) |
Time Since Diagnosis of Metastatic Disease, Median Months (IQR) | 24.97 (6.08, 59.24) | 22.97 (10.78, 45.37) |
Note: Patients could select more than one race
3.2. Intervention Delivery
Of those assigned to the intervention, 55.7% (34/61) completed all five structured PC visits, with an additional 27.9% (17/61) completing as many visits as possible before death or transfer of care (Figure 1). The ten remaining patients either did not initiate (n=5, 8.2%) or complete the intervention (n=5, 8.2%). Within the usual care group, 22% (13/59) of patients had at least one PC visit (range: 1–4) by 24 weeks.
PC clinicians completed 216 surveys detailing the content of their intervention visits. During these visits, 44.3% (27/61) of patients had at least one joint visit with their PC clinician and oncologist. Figure 2 shows the proportion of intervention patients whose PC clinicians discussed the pre-specified topics across the five structured visits. Most PC visits occurred in the clinic, with 4.9% (3/61) of intervention patients having at least one telephone visit (range: 1–2). Finally, 34.4% (21/61) of intervention patients had a median of five additional unstructured PC visits (range: 1–15).
Figure 2.
Clinician-Rated Adherence to Pre-Specified Topics across Five Structured PC Visits
Note: Percentages represent the proportion of intervention patients whose palliative care clinicians discussed the pre-specified intervention topics for each structured visit per the study protocol. The sample sizes for the study visits decreased over time due to patient death, withdrawal, transfer of care, and lost to follow up (Visit 1 n=56, Visit 2 n=49, Visit 3 n=42, Visit 4 n=38, Visit 5 n=33).
3.3. Documentation and Discussion about EOL Care
Per the NLP query (see Table 2 for key terms), intervention patients had higher rates of documented EOL care discussions (67.2% [41/61] vs 40.7% [24/59], p=0.006) and completion of MOLST forms (39.3% [24/61] vs 13.6% [8/59], p=0.002) versus usual care patients. Controlling for age, differences in documented EOL care discussions (OR=2.92; 95%CI=1.38, 6.18; p=0.005) and MOLST form completion (OR=4.09; 95%CI=1.65, 10.14; p=0.002) remained significant. The time from enrollment to documentation of EOL care discussions was shorter for intervention patients versus usual care patients (Mean days=165.63 [SD=22.75] vs 316.75 [SD=45.55], Adjusted B=−152.38, 95%CI=−244.53, −60.23, p=0.002).
Table 2.
Key Terms Used in NLP Method To Query the Electronic Health Record
Domain | Key Terms |
---|---|
Hospice | hospice |
Advance Care Planning/EOL Care | end of life, end-of-life, EOL, end-of-life care, dying, death , comfort care, comfort approach, comfort directed care, comfort measures, cmo, advance care plans/goals, advance care planning, acp, advanced care planning |
Limitation of Life-sustaining Treatment | dnr, dnrdni, dni, dnr/dni, do not resuscitate, do-not-resuscitate, do not intubate, do- not-intubate, no intubation, no mechanical ventilation, no ventilation, no CPR, declines CPR, no cardiopulmonary resuscitation, chest compressions, no defibrillation, no dialysis, no NIPPV, no bipap, no endotracheal intubation, no mechanical intubation |
Additionally, a greater proportion of intervention patients reported discussing with their clinicians any particular wishes about the care they would want to receive if dying versus usual care patients (38.5% [20/52] vs 21.4% [12/56], Adjusted OR=3.10, 95%CI=1.21, 7.94, p=0.019). Yet, we observed no significant differences between study groups regarding the proportion of patients who reported that their cancer is curable (Intervention=16.0% [8/50] vs Usual Care=17.9% [10/56], Adjusted OR=0.37, 95%CI=0.04, 3.78, p=0.400).
3.4. QOL and Mood Symptoms
Analyses of patient-reported outcomes using the terminal-decline joint modeling showed no significant differences in ratings of patient-reported QOL or symptoms of anxiety and depression over time (Table 3). Table 4 details the descriptive statistics for these outcomes at each assessment time point.
Table 3.
Group Differences in Patient-Reported Outcomes across All Time Points Using Terminal Decline Modeling
Four Months Prior to Death* | Six Months Prior to Death* | |||||
---|---|---|---|---|---|---|
Mean | 95%CI | P | Mean | 95%CI | P | |
Usual Care | 95.11 | 91.54, 98.68 | 96.48 | 93.15, 99.81 | ||
Usual Care | 6.23 | 5.35, 7.12 | 6.23 | 5.41, 7.04 | ||
Usual Care | 5.74 | 5.06, 6.42 | 5.51 | 4.88. 6.13 |
Note: We estimated terminal decline and survival distributions with semiparametric models to analyze patient-reported outcomes across all study time points, comparing QOL and symptoms of anxiety and depression between study groups at four and six months prior to death (or prior to date of last follow-up or data cutoff for analysis among those alive). All terminal decline modeling analyses include the patient-reported assessments through week 24 and control for patient age and baseline scores of the outcome variables of interest. FACT-B= Functional Assessment of Cancer Therapy-Breast; HADS=Hospital Anxiety and Depression Scale
Table 4.
Descriptive Statistics for Patient-Reported Outcome Measures across Time Points
Variable* | Usual Care M (SD) | Intervention M (SD) |
---|---|---|
HADS-Depression Subscale | 5.47 (3.57) | 5.16 (3.59) |
HADS-Depression Subscale | 5.13 (3.21) | 5.37 (3.70) |
HADS-Depression Subscale | 5.04 (3.49) | 5.67 (4.12) |
HADS-Depression Subscale | 5.91 (3.60) | 5.75 (4.43) |
HADS-Depression Subscale | 5.31 (3.83) | 4.83 (3.71) |
Note: FACT-B=Functional Assessment of Cancer Therapy-Breast; HADS=Hospital Anxiety and Depression Scale
3.5. Hospice Utilization
Among the patients who died by the data cutoff date, a greater proportion of intervention patients utilized hospice services versus usual care patients (88.6% [31/35] vs 65.5% [19/29], p=0.035). Controlling for age, this difference remained significant (Adjusted OR=4.03, 95%CI=1.10, 14.73, p=0.035). The median days from hospice referral to death was 16.5 (range: 1–243) for intervention patients versus 15.0 (range: 1–228) for usual care patients (p=0.511).
4. Discussion
Patients with MBC remain at risk for worse EOL outcomes despite advances in cancer therapies prolonging survival in this population.2,9,10,20 To address this concern, we conducted a randomized trial of a population-specific PC intervention, which led to higher rates of discussions regarding EOL care preferences between patients and their clinicians. Compared to the control group, a greater proportion of patients with MBC assigned to the intervention reported discussing with clinicians any wishes about the care they would want to receive if dying, had these discussions documented in the EHR, and completed a MOLST form. The intervention also resulted in documentation of EOL care discussions closer to the time of enrollment and higher rates of hospice use compared to the usual care group. The study findings demonstrate that providing a PC intervention later in the illness course is beneficial for supporting women with MBC by not only improving patient-clinician discussions but also enhancing the quality of EOL care.
Only one prior study evaluated a PC model specifically for patients with MBC in which PC clinicians were embedded in a breast oncology clinic.37 This embedded model of care was feasible and acceptable to patients and oncologists as well as associated with increased referrals to PC and fewer ICU stays in the last month of life, as compared to an earlier 24-month period of stand-alone PC.37 To our knowledge, our study advances the field as the first randomized trial to demonstrate improved discussion and documentation of EOL care as well as hospice utilization with a PC intervention integrated within a breast oncology clinic for patients with MBC.
Interestingly, the mean time to documentation of discussions about EOL care was greater than five months in the intervention group, underscoring the need to build patient-clinician trust over time in order to facilitate effective discussions with patients regarding these concerns. Moreover, while the number of intervention visits was fewer than those of prior studies of early integrated PC, questions regarding the scalability of implementing five structured visits remain, especially given that approximately one-third of intervention patients received additional unstructured PC visits, which may have affected outcomes.
Patients with advanced cancer who have EOL discussions with their clinicians are more likely to receive EOL care that is consistent with their wishes and less intensive, which is associated with better patient QOL and caregiver bereavement.4,38,39 Still, clinicians are often hesitant to discuss such topics due to concerns that patients will become distressed.40,41 In our study, we did not observe worse QOL or mood symptoms with the higher rates of EOL care discussions in the intervention versus the usual care group. Yet, the timing of the patient-reported assessments was anchored to enrollment with predefined intervals that did not necessarily correspond to when EOL care discussions occured. We also did not find that intervention patients reported improvements in QOL or mood symptoms relative to usual care patients. These results are inconsistent with prior trials of early, integrated PC for patients with advanced lung cancer who have reported improved QOL and depression symptoms.12,13 The null findings may be related to when the intervention was delivered in the course of disease or because the sample did not report expected decrements in QOL and mood over time often seen in patients with poor-prognosis cancers.
Identifying subgroups of patients with advanced cancer who have unmet PC needs remains a challenge.42 For those with longer disease trajectories, such as patients with MBC, early engagement with PC from the time of diagnosis may be neither clinically indicated nor an efficient use of limited resources. Despite drawing on expert recommendations from breast oncology clinicians to identify patients who have clinical indicators of poor prognosis, more than 40% of our sample was still alive approximately seven months after closing study enrollment, suggesting wide variability in survival. As EHR documentation data are often limited in quality and comprehensiveness, future work ought to consider ways for enhancing computer-assisted tools, such as deep NLP, not only to assist in identifying patients at risk for unmet PC needs but also to specify EOL care process measures and patient-centered outcomes.29,43,44
Several limitations of the study warrant consideration. First, as a single-site trial at a comprehensive cancer center, study findings may not apply to other care settings with more diverse patient populations. Moreover, 22% of the usual care group had at least one PC consultation by 24 weeks, potentially diminishing intervention effects. We did not assess caregiver outcomes, which should be included in future research. Finally, excluding patients already receiving or needing immediate PC at the time of enrollment may have biased the sample to include healthier individuals and limited the capacity of the intervention to improve QOL and mood symptoms.
This study demonstrates the beneficial role of ambulatory PC when targeted to the needs of a specific population, such as those with MBC. Tailored PC interventions will become increasingly necessary given clinician workforce shortages and progress in novel cancer therapeutics prolonging survival in many subgroups of patients diagnosed with metastatic disease. Thus, this study represents an important step forward in enhancing access to essential supportive care services in a timely manner and for improving the quality of EOL care.
5. Acknowledgments
This work was supported by the National Comprehensive Cancer Network/Pfizer Independent Grants for Learning & Change (PI: Temel).
Appendix 1. Intervention Topics for the Five Structured Palliative Care Visits
- Visit 1
- Rapport Building: introducing palliative care, understanding the patient and caregiver experience, and building trust with the patient and caregiver
- Visits 2–3
- Symptom Management: preparing patients for symptoms, assessing and treating symptoms, including referrals to other clinicians, and coordinating management with oncology
- Illness Understanding: exploring goals and values, assessing and informing patient expectations of prognosis and illness process, and communicating with caregivers about illness understanding
- Coping: reviewing and validating prior coping efforts, discussing and advocating for different methods of coping, and supporting caregiver coping
- Visits 4–5
- Treatment Decision-Making: assessing patient goals and values in treatment decision making, discussing treatment considerations, and supporting treatment decisions
- EOL & Advance Care Planning: discussing EOL care options, advance care planning, and supporting caregivers
Appendix 2. Proportion of Patients by Eligibility Criteria Indicating Poor Prognosis
Leptomeningeal disease, n=3 (2.5%)
Progressive brain metastasis after initial radiation therapy, n=7 (5.8%)
Brain metastases and starting whole brain radiation, n=3 (2.5%)
Discharged after an unplanned hospital admission, n=38 (31.7%)
Triple negative disease and starting second-line chemotherapy, n=8 (6.7%)
Received at least three different treatment regimens in a 12-month period, n=32 (26.7%)
HER2+ or ER+ disease and starting third-line therapy, n=4 (3.3%)
Beginning a treatment clinical trial, n=25 (20.8%)
6. References
- 1.Institute of Medicine. Dying in America: Improving Quality and Honoring Individual Preferences Near the End of Life. Washington (DC); 2015. [DOI] [PubMed] [Google Scholar]
- 2.O’Connor TL, Ngamphaiboon N, Groman A, et al. Hospice utilization and end-of-life care in metastatic breast cancer patients at a comprehensive cancer center. J Palliat Med 2015; 18(1): 50–5. [DOI] [PubMed] [Google Scholar]
- 3.Temel JS, Greer JA, Admane S, et al. Code status documentation in the outpatient electronic medical records of patients with metastatic cancer. J Gen Intern Med 2010; 25(2): 150–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wright AA, Zhang B, Ray A, et al. Associations between end-of-life discussions, patient mental health, medical care near death, and caregiver bereavement adjustment. JAMA 2008; 300(14): 1665–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Earle CC, Landrum MB, Souza JM, Neville BA, Weeks JC, Ayanian JZ. Aggressiveness of cancer care near the end of life: is it a quality-of-care issue? J Clin Oncol 2008; 26(23): 3860–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Keating NL, Landrum MB, Guadagnoli E, Winer EP, Ayanian JZ. Care in the months before death and hospice enrollment among older women with advanced breast cancer. J Gen Intern Med 2008; 23(1): 11–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Accordino MK, Wright JD, Vasan S, et al. Association between survival time with metastatic breast cancer and aggressive end-of-life care. Breast Cancer Res Treat 2017; 166(2): 549–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Braga S, Miranda A, Fonseca R, et al. The aggressiveness of cancer care in the last three months of life: a retrospective single centre analysis. Psychooncology 2007; 16(9): 863–8. [DOI] [PubMed] [Google Scholar]
- 9.Mathew A, Achkar T, Abberbock S, et al. Prevalence and determinants of end-of-life chemotherapy use in patients with metastatic breast cancer. Breast J 2017; 23(6): 718–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Shin JA, Parkes A, El-Jawahri A, et al. Retrospective evaluation of palliative care and hospice utilization in hospitalized patients with metastatic breast cancer. Palliat Med 2016; 30(9): 854–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gagnon B, Mayo NE, Hanley J, MacDonald N. Pattern of care at the end of life: does age make a difference in what happens to women with breast cancer? J Clin Oncol 2004; 22(17): 3458–65. [DOI] [PubMed] [Google Scholar]
- 12.Temel JS, Greer JA, Muzikansky A, et al. Early palliative care for patients with metastatic non-small-cell lung cancer. N Engl J Med 2010; 363(8): 733–42. [DOI] [PubMed] [Google Scholar]
- 13.Temel JS, Greer JA, El-Jawahri A, et al. Effects of Early Integrated Palliative Care in Patients With Lung and GI Cancer: A Randomized Clinical Trial. J Clin Oncol 2017; 35(8): 834–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Temel JS, Greer JA, Admane S, et al. Longitudinal perceptions of prognosis and goals of therapy in patients with metastatic non-small-cell lung cancer: results of a randomized study of early palliative care. J Clin Oncol 2011; 29(17): 2319–26. [DOI] [PubMed] [Google Scholar]
- 15.Greer JA, Pirl WF, Jackson VA, et al. Effect of early palliative care on chemotherapy use and end-of-life care in patients with metastatic non-small-cell lung cancer. J Clin Oncol 2012; 30(4): 394–400. [DOI] [PubMed] [Google Scholar]
- 16.Thomas TH, Jackson VA, Carlson H, et al. Communication Differences between Oncologists and Palliative Care Clinicians: A Qualitative Analysis of Early, Integrated Palliative Care in Patients with Advanced Cancer. J Palliat Med 2019; 22(1): 41–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hoerger M, Greer JA, Jackson VA, et al. Defining the Elements of Early Palliative Care That Are Associated With Patient-Reported Outcomes and the Delivery of End-of-Life Care. J Clin Oncol 2018; 36(11): 1096–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ferrell BR, Temel JS, Temin S, et al. Integration of Palliative Care Into Standard Oncology Care: American Society of Clinical Oncology Clinical Practice Guideline Update. J Clin Oncol 2017; 35(1): 96–112. [DOI] [PubMed] [Google Scholar]
- 19.Levy M, Smith T, Alvarez-Perez A, et al. Palliative Care Version 1.2016. J Natl Compr Canc Netw 2016; 14(1): 82–113. [DOI] [PubMed] [Google Scholar]
- 20.Caswell-Jin JL, Plevritis SK, Tian L, et al. Change in Survival in Metastatic Breast Cancer with Treatment Advances: Meta-Analysis and Systematic Review. JNCI Cancer Spectr 2018; 2(4): pky062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hui D, Mori M, Watanabe SM, et al. Referral criteria for outpatient specialty palliative cancer care: an international consensus. Lancet Oncol 2016; 17(12): e552–e9. [DOI] [PubMed] [Google Scholar]
- 22.Hui D, Meng YC, Bruera S, et al. Referral Criteria for Outpatient Palliative Cancer Care: A Systematic Review. Oncologist 2016; 21(7): 895–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Bakitas M, Lyons KD, Hegel MT, et al. Effects of a palliative care intervention on clinical outcomes in patients with advanced cancer: the Project ENABLE II randomized controlled trial. JAMA 2009; 302(7): 741–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bakitas MA, Tosteson TD, Li Z, et al. Early Versus Delayed Initiation of Concurrent Palliative Oncology Care: Patient Outcomes in the ENABLE III Randomized Controlled Trial. J Clin Oncol 2015; 33(13): 1438–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zimmermann C, Swami N, Krzyzanowska M, et al. Early palliative care for patients with advanced cancer: a cluster-randomised controlled trial. Lancet 2014; 383(9930): 1721–30. [DOI] [PubMed] [Google Scholar]
- 26.Haun MW, Estel S, Rucker G, et al. Early palliative care for adults with advanced cancer. Cochrane Database Syst Rev 2017; 6: CD011129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Vanbutsele G, Pardon K, Van Belle S, et al. Effect of early and systematic integration of palliative care in patients with advanced cancer: a randomised controlled trial. Lancet Oncol 2018; 19(3): 394–404. [DOI] [PubMed] [Google Scholar]
- 28.Yoong J, Park ER, Greer JA, et al. Early palliative care in advanced lung cancer: a qualitative study. JAMA Intern Med 2013; 173(4): 283–90. [DOI] [PubMed] [Google Scholar]
- 29.Brizzi K, Zupanc SN, Udelsman BV, et al. Natural Language Processing to Assess Palliative Care and End-of-Life Process Measures in Patients With Breast Cancer With Leptomeningeal Disease. Am J Hosp Palliat Care 2019: 1049909119885585. [DOI] [PubMed] [Google Scholar]
- 30.Lilley EJ, Lindvall C, Lillemoe KD, Tulsky JA, Wiener DC, Cooper Z. Measuring Processes of Care in Palliative Surgery: A Novel Approach Using Natural Language Processing. Ann Surg 2018; 267(5): 823–5. [DOI] [PubMed] [Google Scholar]
- 31.Lindvall C, Lilley EJ, Zupanc SN, et al. Natural Language Processing to Assess End-of-Life Quality Indicators in Cancer Patients Receiving Palliative Surgery. J Palliat Med 2019; 22(2): 183–7. [DOI] [PubMed] [Google Scholar]
- 32.Chan A, Chien I, Moseley E, et al. Deep learning algorithms to identify documentation of serious illness conversations during intensive care unit admissions. Palliat Med 2019; 33(2): 187–96. [DOI] [PubMed] [Google Scholar]
- 33.Brady MJ, Cella DF, Mo F, et al. Reliability and validity of the Functional Assessment of Cancer Therapy-Breast quality-of-life instrument. J Clin Oncol 1997; 15(3): 974–86. [DOI] [PubMed] [Google Scholar]
- 34.Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983; 67(6): 361–70. [DOI] [PubMed] [Google Scholar]
- 35.Bjelland I, Dahl AA, Haug TT, Neckelmann D. The validity of the Hospital Anxiety and Depression Scale. An updated literature review. J Psychosom Res 2002; 52(2): 69–77. [DOI] [PubMed] [Google Scholar]
- 36.Li Z, Tosteson TD, Bakitas MA. Joint modeling quality of life and survival using a terminal decline model in palliative care studies. Stat Med 2013; 32(8): 1394–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Rabow M, Small R, Jow A, et al. The value of embedding: integrated palliative care for patients with metastatic breast cancer. Breast Cancer Res Treat 2018; 167(3): 703–8. [DOI] [PubMed] [Google Scholar]
- 38.Mack JW, Weeks JC, Wright AA, Block SD, Prigerson HG. End-of-life discussions, goal attainment, and distress at the end of life: predictors and outcomes of receipt of care consistent with preferences. J Clin Oncol 2010; 28(7): 1203–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Wright AA, Keating NL, Ayanian JZ, et al. Family Perspectives on Aggressive Cancer Care Near the End of Life. JAMA 2016; 315(3): 284–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Mack JW, Smith TJ. Reasons why physicians do not have discussions about poor prognosis, why it matters, and what can be improved. J Clin Oncol 2012; 30(22): 2715–7. [DOI] [PubMed] [Google Scholar]
- 41.Norals TE, Smith TJ. Advance Care Planning Discussions: Why They Should Happen, Why They Don’t, and How We Can Facilitate the Process. Oncology (Williston Park) 2015; 29(8): 567–71. [PubMed] [Google Scholar]
- 42.Collins A, Sundararajan V, Burchell J, et al. Transition Points for the Routine Integration of Palliative Care in Patients With Advanced Cancer. J Pain Symptom Manage 2018; 56(2): 185–94. [DOI] [PubMed] [Google Scholar]
- 43.Forsyth AW, Barzilay R, Hughes KS, et al. Machine Learning Methods to Extract Documentation of Breast Cancer Symptoms From Electronic Health Records. J Pain Symptom Manage 2018; 55(6): 1492–9. [DOI] [PubMed] [Google Scholar]
- 44.Parikh RB, Manz C, Chivers C, et al. Machine Learning Approaches to Predict 6-Month Mortality Among Patients With Cancer. JAMA Netw Open 2019; 2(10): e1915997. [DOI] [PMC free article] [PubMed] [Google Scholar]