Phase 1 trials offer patients with advanced cancer the opportunity to pursue life-prolonging cancer treatments. In the current study, the timing of referral and symptom burden between patients referred to palliative care by phase 1 oncologists and those referred by non-phase 1 oncologists were compared.
All 57 patients with advanced solid tumors who were referred by phase 1 oncologists to the palliative care outpatient clinic at The University of Texas M. D. Anderson Cancer Center (MDACC) between 2007 and 2008 were included. The comparison cohort was comprised of 114 non-phase 1 patients who were stratified by age, sex, and cancer diagnosis in a 1:2 ratio. Information regarding patient characteristics, Edmonton Symptom Assessment Scale (ESAS), timing of referral, and survival was retrieved.
Both cohorts had the following matched characteristics: average age of 57 years, with44% of the patients being female and 47% having gastrointestinal cancers. At the time of palliative care consultation, patients referred by phase 1 oncologists were more likely than patients referred by non-phase 1 oncologists to have a better performance status (Eastern Cooperative Oncology Group 0–1: 61% vs 36% [P = .003). The ESAS was not significantly different with the exception of better well-being in the phase 1 cohort (mean, 4.5 vs 5.5; P = .03). No difference was found for the duration between registration at MDACC and palliative care consultation (13 months vs 11 months; P = .41) and overall survival from the time of palliative care consultation (5 months vs 4 months; P = .69).
Patients with advanced cancer usually develop a higher symptom burden with disease progression, and have fewer standard treatment options available.1, 2 A small percentage of these patients enroll in phase 1 clinical trials, fueled by the hopes of better disease control and improved survival.3 Although phase 1 agents may result in symptom benefit through disease stabilization or shrinkage, they can also be associated with various toxicities and logistical burden, including frequent hospital visits and investigations.4, 5 These, coupled with cancer-related symptoms and complications, can significantly compromise a patient's quality of life.
Phase 1 patients generally have a poor prognosis, with a median survival of 6 to 9 months.6-8 Given the significant morbidity and mortality reported among phase 1 patients, the involvement of palliative care clinicians can be potentially beneficial. Palliative care has evolved as a discipline that focuses on the improvement of the quality of life of cancer patients and their families through the early identification, assessment, and treatment of symptoms.9 Timely referral to palliative care is an important indicator of quality of care because patients gain access to multidimensional care early in the trajectory of illness.10, 11 However, studies have indicated that phase 1 patients are less likely to consider palliative care, home health aides, counselors, and chaplains.12 Thus, the pursuit of phase 1 therapy may potentially delay referral to palliative care.
To our knowledge, only 1 study concerning the symptom burden of phase 1 patients compared with non-phase 1 patients has been published to date,12 and no information regarding the impact of phase 1 involvement on the timing of referral to palliative care is available. A better understanding of the timing of palliative care referral and symptom profile can provide the foundation for optimizing care for these individuals. Using a retrospective cohort design, we compared the timing of referral and symptom burden between patients referred to palliative care by phase 1 oncologists and those referred by non-phase 1 oncologists.
MATERIALS AND METHODS
The Institutional Review Board at The University of Texas M. D. Anderson Cancer Center (MDACC) approved the current study and waived the requirement for informed consent. All patients with advanced solid tumors who were referred to the outpatient supportive care clinic at MDACC between January 1, 2007 and December 31, 2008 as the first palliative care consultation were identified. Fifty-seven patients were referred by phase 1 oncologists, and were designated as the phase 1 cohort. Among the remaining patients, we randomly selected 114 as the non-phase 1 cohort, and these patients were stratified by age (<60 years or ≥60 years), sex (male or female), and cancer diagnosis (breast, gastrointestinal, genitourinary, gynecologic, head and neck, lung, and other) in a 1:2 ratio.
Patient Characteristics, Symptoms, and Timing of Referral
We retrospectively retrieved patient demographics (age, sex, and race); cancer diagnosis; timing of diagnosis; encounters with medical oncology, phase 1, and palliative care; treatment history; and survival from institutional databases, electronic health records, and the Tumor Registry vital statistics database. We also collected information regarding the Edmonton Symptom Assessment Scale (ESAS), the Memorial Delirium Assessment Scale (MDAS), and the Eastern Cooperative Oncology Group (ECOG) performance status at the time of palliative care outpatient consultation.
We summarized the baseline demographics and symptom profile using descriptive statistics, including medians, means, standard deviations, ranges, and frequencies together with 95% confidence intervals (95% CIs).
We compared the baseline characteristics and symptom profile between the phase 1 and non-phase 1 patient cohorts. Comparisons were made using the Student t test for continuous variables that were normally distributed (ie, ESAS); the Mann-Whitney U test for continuous, nonparametric variables (eg, prior systemic therapy courses); and the chi-square test or Fisher exact test for categorical variables (eg, MDAS). A 2-sided P value <.05 was considered to be statistically significant.
Timing of palliative care referral was defined a priori based on 2 intervals, including time from registration at MDACC to palliative care consultation and overall survival from the time of palliative care consultation. All time-event analyses were plotted using the Kaplan-Meier method, and survival curves were compared with the log-rank test.13, 14 Overall survival was calculated from the date of palliative care referral to the date of death from any cause or the date at which the patient was last known to be alive. Multivariate analysis was performed using the Cox proportional hazards model with backward elimination.
The Statistical Package for the Social Sciences (SPSS) (version 16.0; SPSS Inc, Chicago, Ill) software was used for statistical analysis.
Table 1 summarizes patient demographics at the time of the first outpatient palliative care clinic consultation. Patients referred by phase 1 oncologists were less likely to present with metastatic disease and to have received more prior chemotherapy regimens compared with those referred by non-phase 1 oncologists. Among the phase 1 patient cohort, 6 (11%), 13 (23%), 13 (23%), and 25 (43%) patients completed 1, 2, 3, and ≥4 lines of chemotherapy, respectively, before palliative care consultation compared with 33 (29%), 25 (22%), 14 (12%), and 19 (17%) patients, respectively, in the non-phase 1 group. A small percentage (26%) of patients in the phase 1 cohort were seen directly by phase 1 without medical oncology involvement from our institution. Among the non-phase 1 cohort, 12 of 114 (11%) patients had also been seen during phase 1, but were referred by nononcologists to palliative care.
Table 1. Baseline Characteristics at Time of Palliative Care Clinic Consultation
Phase 1 Referrals (N=57)
Non-Phase 1 Referrals (N=114)
SD indicates standard deviation; 95% CI, 95% confidence interval.
Between the time of advanced cancer diagnosis and palliative care consultation.
Denotes the time sequence of patient encounters with oncology, phase 1, and palliative care, ordered by the date of first visit to each service.
Age (SD), y
Head and neck
No. of lines of chemotherapy regimens prior to palliative care consultationa
Median interval from diagnosis of advanced cancer to palliative care consultation (95% CI), mo
The symptom burden at the time of the first outpatient palliative care clinic consultation is shown in Table 2. Patients referred by phase 1 oncologists had better ECOG performance status and were less likely to be delirious (MDAS ≥8) at the time of presentation. The 2 cohorts had similar physical and psychosocial symptoms as assessed by ESAS, although the phase 1 cohort had better overall well-being (mean, 4.5 vs 5.5; P = .03) compared with the non-phase 1 cohort.
Table 2. Symptom Burden at Time of Palliative Care Clinic Consultation
Phase 1 Referrals (N=57)
Non-Phase 1 Referrals (N=114)
SD indicates standard deviation; ECOG, Eastern Cooperative Oncology Group; MDAS, Memorial Delirium Assessment Scale.
Mean Edmonton Symptom Assessment Scale (SD)
ECOG performance status
Delirium (MDAS ≥8)
Medical Oncology and Phase 1 Involvement
Compared with the non-phase 1 cohort, patients referred by phase 1 oncologists had significantly longer survival from the time of diagnosis of advanced cancer (Table 3). Both cohorts were followed by medical oncology for a similar duration (17 months vs 14 months; P = .36). Patients in the phase 1 cohort were referred to phase 1 oncologists at a median of 23 months (95% CI, 11-45 months) from the time of diagnosis of advanced cancer, and a median of 6.5 months (95% CI, 4.4-8.6 months) before death.
Table 3. Timing of Palliative Care and Phase 1 Referrals
Phase 1 Referrals (N=57)
Non-Phase 1 Referrals (N=114)
95% CI indicates 95% confidence interval; MDACC, The University of Texas M. D. Anderson Cancer Center; IQR, interquartile range.
Median interval (95% CI), mo
Diagnosis of advanced cancer to death
Diagnosis of advanced cancer to palliative care consultation
MDACC registration to palliative care consultation
Palliative care consultation to death
First medical oncology contact to palliative care consultation
First referring team contact to palliative care consultation
Median no. of clinic visits (IQR)
Medical oncology visits before palliative care consultation
Palliative care visits until last follow-up/death
No difference was found with regard to survival from the referring team's last visit (median, 28 days vs 41 days; P = .84); however, the last medical oncology visit was found to have happened earlier in the phase 1 cohort (200 days vs 41 days; P < .001), suggesting that phase 1 oncologists, rather than the referring oncologists, were primarily responsible for the care of these patients in their last few months of life.
Quality of End-of-Life Care
Among all the patients who died, those referred by phase 1 oncologists were more likely than the non-phase 1 cohort to have received chemotherapy within the last 30 days of life (31% vs 13%; P = .014). The interval between last chemotherapy and death also appeared to be shorter for the phase 1 cohort (median, 60 days vs 81 days; P = .06), although this did not achieve statistical significance.
Only a small percentage of patients in both cohorts were admitted to the intensive care unit within the last 30 days of life (4% vs 6%; P >.99). We did not detect any differences in the percentage of in-hospital deaths between the 2 groups (18% vs 28%; P = .21).
Palliative Care Referral
Patients referred by phase 1 oncologists were found to have a significantly longer interval from the time of diagnosis of advanced cancer to palliative care consultation, as well as a longer interval between first oncology contact and palliative care consultation (Table 3). However, the 2 key markers of timing of palliative care referral (time from MDACC registration to palliative care consultation and overall survival from time of palliative care consultation) did not differ significantly between the 2 groups (Table 3). On Cox regression multivariate analysis, only cancer diagnosis was found to be significantly associated with the timing of palliative care referral (Table 4).
Table 4. Cox Regression Multivariate Analysis of Survival from Palliative Care Consultationa
Variables included in this model were age, sex, race, cancer diagnosis, stage, referring team, ECOG performance status at presentation, number of medical oncology visits before palliative care consultation, and number of chemotherapy courses before palliative care consultation. Because this model aimed to identify factors that affect the timing of referral rather than factors that confer a poor prognosis, we specifically excluded symptoms from the analysis because some are known to be prognostic factors (eg, delirium and dyspnea) that could potentially confound this analysis.
Included all cancer types other than breast, gastrointestinal, genitourinary, gynecologic, and head and neck cancers. Examples included sarcoma as well as neurologic and endocrine tumors.
The duration between the referring team's initial contact with the patient and palliative care referral was shorter for patients in the phase 1 cohort, suggesting that these patients were referred promptly to palliative care (Table 3). No difference was found with regard to the duration of overlap between palliative care consultation and the referring team's last visit (median, 63 days vs 44 days; P = .84) (Fig. 1).
Consistent with these findings, the number of medical oncology clinic visits before palliative care consultation and the number of palliative care clinic visits until death were found to be similar between the 2 cohorts (Table 3). Patients in the phase 1 cohort had a median of 4 phase 1 visits (interquartile range, 1.5-10.5 visits) before a palliative care consultation.
In this retrospective cohort study, we found that patients referred to palliative care by phase 1 oncologists had a better ECOG performance status but similar symptom burden compared with those referred by non-phase 1 oncologists, and were more likely to receive chemotherapy close to the end of their life. Survival from the time of palliative care referral was approximately 4 months for both cohorts, suggesting that phase 1 involvement did not delay palliative care referral. This supports the development of a successful simultaneous care model.
One interesting, although not completely surprising, finding from the current study was that the phase 1 cohort had a longer overall survival from the time of diagnosis of advanced cancer compared with the non-phase 1 cohort, suggesting that these patients generally had a less aggressive disease course compared with patients who were not referred.15 In contrast, several patients in the non-phase 1 cohort deteriorated rapidly, thereby limiting their ability to enroll in clinical trials.
We found that patients from the phase 1 cohort had a better ECOG performance status and were less likely to be delirious at the time of presentation to palliative care; this finding is not surprising given that patients need to be well enough to be referred for phase 1 treatments.15, 16 We found that phase 1 patients otherwise had similar symptom expression compared with non-phase 1 patients. These findings are consistent with a recent prospective cohort study from a different institution using the Memorial Symptom Assessment Scale.12 Similar to other studies regarding palliative care populations, fatigue, pain, and anorexia were found to be the most common and severe symptoms in both cohorts.17 Importantly, our results support that both phase 1 and non-phase 1 patients have significant physical and psychological distress at the time of presentation to the outpatient palliative care clinic, indicating the need for intensive interventions.
We had several reasons to suspect that phase 1 patients could be referred to palliative care later than non-phase 1 patients. Phase 1 treatments are typically offered late in the disease trajectory, when patients have already exhausted all standard treatments. Thus, although the non-phase 1 oncologists have ample opportunities to make palliative care referrals early in the disease course, the timing of referral by phase 1 physicians depends on when the phase 1 consultations take place, which are generally within the last few months of life.
In addition to logistical challenges, the literature suggests that patients taking part in phase 1 trials are not psychologically prepared for the transition to end-of-life care.12, 18 Agrawal et al surveyed 163 patients on phase 1 protocols regarding their decision-making process. Although >80% of patients stated that they were aware of palliative care and hospice as alternatives to phase 1 trials, fewer than 10% seriously considered these options for themselves.18 Potential explanations include a lack of understanding of palliative care services, lack of a self-perceived need for support services because of better performance status, and the desire to focus only on cancer treatments. Indeed, patients embarking on phase 1 trials generally have heightened expectations with regard to survival and treatment outcomes.19, 20 This, coupled with a sense of denial, makes them much less likely to desire palliative care services. Finally, the misconception by some oncologists that a palliative care referral could destroy hope presents another barrier to the referral process.21, 22
Despite the many factors that could potentially limit palliative care referral among phase 1 patients, we were encouraged to find that phase 1 involvement did not appear to delay palliative care referral. In the current study, the median duration between first phase 1 contact and palliative care referral was only 31 days, suggesting that phase 1 physicians made their referral relatively quickly, without significant delays. Because phase 1 physicians work closely with patients with advanced cancer near the end of life, they may have a better understanding of the palliative care needs of their patients, and the potential benefits related to a timely referral. For individuals too sick to enroll on an experimental protocol, they may be referred to palliative care for transition to end-of-life care. For other patients participating in phase 1 trials, palliative care helps to support them through treatment by optimizing their functional status, symptoms, and support systems.
Consistent with this model of integrated care, patients referred by phase 1 oncologists were more likely to receive chemotherapy within the last 30 days of life. This finding is not unexpected given that these patients were actively seeking further treatments. The administration of chemotherapy close to the end of life has been used as an indicator of poor quality of cancer care.23-25 However, this marker is of limited value as an outcome measure because the timing of death cannot be accurately predicted. There are always patients who desire aggressive treatments despite understanding the unfavorable risk:benefit ratio. Finally, the emergence of targeted therapies with a lower toxicity profile compared with traditional cytotoxic therapies has allowed sicker patients to receive anticancer treatments later in the disease trajectory. Thus, the 30-day criterion may not be appropriate for phase 1 cancer patients.
Close collaborations between phase 1 and palliative care physicians under a simultaneous care model can help optimize both quantity and quality of life, while tailoring care to the individual's needs.26-28 Clinicians, patients, and families should understand that they do not have to choose between phase 1 and palliative care, but could take advantage of both services synergistically. Under this integrated care approach, phase 1 oncologists can deliver the latest cancer treatments with the aim of cancer control, while the palliative care team focuses on symptom management, psychosocial interventions, family counseling, and transition of care. Given the significant symptom burden, emotional distress, and poor prognosis of phase 1 patients, nearly all of them may benefit from this simultaneous care approach.
To the best of our knowledge, the optimal timing for palliative care referral has not been defined to date, although the general understanding is that patients are referred late.29-31 The American Society of Clinical Oncology recently published a position statement supporting the vision of integrating palliative care into oncology practice from the time of cancer diagnosis.11 In our study of patients who were first referred as outpatients, palliative care involvement was limited to the last 4 months of life, with a median of 4 clinical encounters. Thus, much work remains to be done to provide palliative care earlier in the disease trajectory, ideally before the phase 1 oncologist visit. Studies from our group and others have demonstrated that palliative care referral is dependent on various factors, including the oncologist's perception and attitudes,21, 32 patient characteristics and preferences,33, 34 and healthcare infrastructure and policies.35 Consistent with the literature,36 we found that a cancer diagnosis, which dictates which team of oncologists the patient sees, is a key determinant regarding the timing of referral to palliative care. Although not examined in the current study, factors associated with referral to phase 1 oncologists may include institutional resources, the oncologist's attitudes, and patient education and preferences, as well as cancer site and stage of disease.15, 37, 38
The current study has several limitations. First, data were collected retrospectively. Second, although our phase 1 and palliative care programs represent some of the largest programs in the United States, the sample size was small. This is partly because we elected to focus on outpatient referrals as a model of integration. Third, the current study only examined the timing of referral because we did not have access to the data of patients who, although seen by phase 1 and non-phase 1 oncologists, were not subsequently referred to palliative care. Fourth, approximately 11% of patients in the non-phase 1 cohort had consulted phase 1 oncologists either before or after palliative care consultation, although they were not referred by the phase 1 oncologist. This “contamination” may have reduced any observed differences in our outcomes. Finally, findings from the current study are institution-specific and are not generalizable to other programs. Nevertheless, it demonstrates how 2 healthcare teams with seemingly opposing objectives can care for patients collaboratively.
To our knowledge, this is the first study to examine the timing of referral to palliative care between patients treated by phase 1 and non-phase 1 oncologists. We found that phase 1 involvement did not delay palliative care referral in our institution. Despite having a better ECOG performance status, patients referred by phase 1 oncologists were found to have a significant symptom burden similar to that of patients referred by non-phase 1 services, and were more likely to be receiving chemotherapy closer to the time of death. These findings indicate the need for interdisciplinary palliative care interventions concurrent with phase 1 under a simultaneous care model. Although not every patient requires intensive symptom management up front, timely referral can facilitate rapport building, longitudinal psychosocial support, early recognition and treatment of symptoms, and advance care planning. To ensure timely access to comprehensive cancer care for all patients, improvements in reimbursement policy; palliative care resources; and the education of oncologists, patients, and families as well as research examining novel integration models are urgently needed.
We would like to thank Ray Chacko for assistance with database management.
CONFLICT OF INTEREST DISCLOSURES
Supported in part by National Institutes of Health grants RO1NR010162-01A1, RO1CA122292-01, and RO1 CA124481-01 (to Dr. Bruera); and the Clinician Investigator Program, Royal College of Physicians and Surgeons of Canada (to Dr. Hui).