INTRODUCTION
Patients with advanced cancer may experience severe physical and psychological symptoms during the course of the disease. (Dodd et al., Reference Dodd, Miaskowski and Paul2001; Paice, Reference Paice2004; Trask & Griffith, Reference Trask and Griffith2004) Symptoms reported in advanced cancer patients include: pain (in 41–76% of patients), fatigue (in 90%), nausea (in 24–90%), depression (in 33–40%), anxiety (in 57–68%), sedation/confusion (in 46–60%), anorexia (in 85%), dyspnea (in 12–58%), sleep disturbance (in 50%), and constipation (in 65%) (Walsh et al., Reference Walsh, Donnelly and Rybicki2000; Klinkenberg et al., Reference Klinkenberg, Willems and van der Wal2004) Many of these symptoms occurred simultaneously as groups or clusters, (Portenoy et al., Reference Portenoy, Thaler and Kornblith1994; Chang et al., Reference Chang, Hwang and Feuerman2000; Cleeland et al., Reference Cleeland, Mendoza and Wang2000) in particular pain, fatigue, depression, and sleep disturbance (Walsh & Rybicki, Reference Walsh and Rybicki2006).
A symptom cluster (SC) is defined as two or more symptoms that co-occur and are strongly correlated among symptoms within that group (Dodd et al., Reference Dodd, Miaskowski and Paul2001; Kim et al., Reference Kim, McGuire and Tulman2005; Miaskowski, Reference Miaskowski2006). The etiology of SCs is not currently known; however, the last decade has seen increased interest in the field of SC research in patients with advanced cancer (Dodd et al., Reference Dodd, Miaskowski and Paul2001). These symptoms may be the result of patient factors, cancer and/or treatment (e.g., cytotoxic chemotherapy, radiation, or surgery) (Rosenthal et al., Reference Rosenthal, Mendoza and Chambers2007; Hadi et al., Reference Hadi, Fan and Hird2008; Cleeland et al., Reference Cleeland, Mendoza and Wang2011; Thong et al., Reference Thong, Mols and Lemmens2011; Yang et al., Reference Yang, Cheville and Wampfler2012) and can have severe psychosocial impact on patients' quality of life (Kramer et al., Reference Kramer, Curran and Piccart2000; Efficace et al., Reference Efficace, Biganzoli and Piccart2004). Even though studies have been undertaken to understand the frequency and intensity of SCs, few studies focused on identifying SCs in patients receiving outpatient palliative care (Cheung et al., Reference Cheung, Le and Zimmermann2009; Tsai et al., Reference Tsai, Wu and Chiu2010). There also are limited data on the consistency and severity of SCs presented following an outpatient interdisciplinary team consultation led by palliative care specialists.
Further knowledge of SCs is needed because patients may receive limited contact with the palliative care team because of late referrals or outpatient status, and these data could be helpful to develop effective management strategies and thus provide quality palliative care to patients with advanced cancer. The aim of this retrospective trial was to determine the consistency and severity of SCs after an outpatient interdisciplinary palliative team consultation.
METHODS
Patient Selection
We reviewed the medical records of consecutive patients with advanced cancer (defined as locally advanced and/or metastatic) seen at the outpatient Supportive Care Center in The University of Texas MD Anderson Cancer Center between January 2003 and October 2008. Advanced cancer patients who had at least one follow-up visit within 7–30 days of initial visits and who completed an Edmonton Symptom Assessment Scale (ESAS) questionnaire at the time of initial and follow-up visits were included in the study. Of the 1378 patients initially enrolled in the trial, 5 were excluded because of missing data from the ESAS at the baseline visit (n = 4) or at follow-up (n = 1).
We received institutional review board approval for conducting this study.
Process of Palliative Care Service
Palliative care in the outpatient setting is provided by an interdisciplinary team led by board-certified palliative care specialists. The team includes a physician and may involve other specialists such as a registered nurse trained specifically in palliative care, a pharmacist, a nutritionist, a chaplain, a social worker, and a palliative- and psychiatry-trained advanced practice nurse who provides counseling services. The mutual coverage by our team helps to maintain consistent case assessment and management as well as communication with patients and their families. Care providers in wound management, speech therapy, occupational therapy, and physical therapy also are consulted when needed. Each patient's care plan is developed according to a standardized management plan (Bruera & Elsayem, Reference Bruera and Elsayem2008).
Once the registered nurse assesses the patient using the ESAS, the Memorial Delirium Assessment Scale (to screen for delirium) and constipation and family support questionnaires, the team proposes a treatment plan on the basis of the input of patients and their families. This plan follows the guidelines of the National Comprehensive Cancer Network and National Consensus Project (Levy et al., Reference Levy, Back and Bazargan2006). These guidelines demonstrate the importance of the multidimensional team approach, including the following aspects of care:
• structure and processes of a treatment plan
• physical, psychosocial and psychiatric
• social and cultural
• ethical and legal
• spiritual, religious, and existential
• the unique needs of the imminently dying patient
Assessments
We collected data on each patient's demographic features including age, sex, and race as well as cancer diagnosis.
ESAS
The ESAS is a self-reported instrument developed and modified to measure symptoms of cancer patients in palliative care (Bruera et al., Reference Bruera, Kuehn and Miller1991). It catalogs pain, fatigue, nausea, depression, anxiety, loss of appetite, drowsiness, well-being, dyspnea, and sleep disturbance. ESAS is validated as a simple bedside method for symptom assessment in patients receiving palliative care (Watanabe et al., Reference Watanabe, Nekolaichuk and Beaumont2011). The symptom distress score for each defined SC as well as total scores were calculated with a sum of score of each item included in that same SC.
Statistical Analysis
We analyzed demographic variables (age, sex, and race) and clinical characteristics of patients using descriptive statistics (means, medians, frequencies, and percentages) and compared differences between baseline and a follow-up ESAS score using a Wilcoxon signed-rank test. We used the Spearman rank correlation coefficient to measure the association between ESAS symptom scores at baseline and at follow-up. Principal component factor analysis was used to determine SCs at baseline and follow-up by transforming the data to the natural log scale to obtain a normal distribution. The analysis of symptoms was based on a patient's ESAS score at the baseline Supportive Care Center visit and the first subsequent follow-up. Similarity between symptoms was measured using correlations, and average linkage was used to join clusters.
RESULTS
Patient characteristics are shown in Table 1. In the 1373 patients in the study, the median age was 59 years and 53% were male (732/1373). The most common cancer was lung cancer (24%) followed by gastrointestinal cancer (20%). The median time between the initial visit and follow-up was 14 days.
Table 1. Patients' characteristics (n = 1373)
All ESAS values improved between baseline and follow-up except for appetite (Table 2).
Table 2. ESAS scores at baseline and follow-up
ESAS, Edmonton Symptom Assessment Scale; SDS, total symptom distress score; SD, standard deviation; Q1–Q3, endpoints of the interquartile range.
Tables 3 and 4 show that all ESAS items strongly correlated with each other (p < 0.0001) except dyspnea and pain (p = 0.30) and drowsiness and pain (p = 0.06) at baseline. The highest correlation was observed between anxiety and depression, both at baseline and follow-up.
Table 3. Spearman's correlation of ESAS scores at baseline
*p < 0.0001 (2-tailed). †p = 0.0.28. Correlation is significant at the 0.05 level (2-tailed).
ap = 0.2.
ESAS, Edmonton Symptom Assessment Scale.
Table 4. Spearman's correlation of ESAS scores at follow-up
*p < 0.0001 (2-tailed). ESAS, Edmonton Symptom Assessment Scale.
At baseline, the principal component factor analysis identified two SCs that accounted for 42% of the variance: physical and psychological (Table 5). The physical SC included pain, fatigue, nausea, loss of appetite, drowsiness, well-being, dyspnea, and sleep disturbance; the standardized Cronbach's α score was 0.68. The psychological SC consisted of depression and anxiety; standardized Cronbach's α score was 0.77. At the follow-up visit, we identified two clusters with 44% variance. Components for each cluster were the same as in the baseline SCs and Cronbach's α scores were 0.71 for the physical SC and 0.80 for the psychological SC.
Table 5. Principal component analysis of baseline and follow-up ESAS scores
*Indicates symptom clustering in each factor. Baseline: On the basis of the eigenvalues, we chose to examine two factors that accounted for 42% of variance. Follow-up: On the basis of the eigenvalues, we chose two factors that explained 44% of the variance. ESAS, Edmonton Symptom Assessment Scale.
We next compared the mean total symptom distress score and the mean symptom score for each SC between baseline and follow-up (Fig. 1). Physical, psychological, and total distress scores all decreased significantly from the time of baseline to follow-up (p < 0.001).
Fig. 1. Mean symptom distress scores at baseline and follow-up for each symptom cluster and the total (*p < 0.001, SE: Standard error).
DISCUSSION
Our data lead us to conclude that SC composition in advanced cancer patients presenting to outpatient palliative care can be easily differentiated into physical and psychological components at the initial visit, and that these two SCs are consistent at follow-up. Our data also indicate that advanced cancer patients in outpatient palliative care who experience severe physical and psychological symptoms at presentation can attain significant improvement by the time of the first follow-up visit after an initial palliative care consultation.
To our knowledge, this is the first study to provide information about the consistency of SCs and changes in the severity of both physical and psychological SCs at the first follow-up visit after an initial palliative consultation. Such a consistent outcome measure may help us target a common treatment strategy that would alleviate the severity of physical or psychological symptoms. This consistency also may enable us to more easily monitor changes in symptom severity at follow-up visits.
Prior studies have examined the consistency of symptom SCs longitudinally in patients with advanced cancer. Some of these studies used ESAS items for SC analysis. In a study by Chow et al. (Reference Chow, Fan and Hadi2008) of 170 patients with brain metastases receiving radiation therapy, three SCs were identified: a fatigue, drowsiness, shortness of breath, and pain cluster; an anxiety and depression cluster; and a poor appetite, nausea, and poor sense of well-being cluster. In contrast with our results, however, over time during radiation therapy, the composition of the SCs changed (Chow et al., Reference Chow, Fan and Hadi2008).
In another study by Chow et al. of 518 patients with bone metastases receiving radiation therapy, three SCs were identified: a fatigue pain, drowsiness, and poor sense of feeling of well-being cluster; an anxiety and depression cluster; and a shortness of breath, nausea, and poor appetite cluster (Chow et al., Reference Chow, Fan and Hadi2007). As in the previous study, over time during radiation therapy, the composition of the SCs changed. Like us, Chow et al. found depression and anxiety to be a distinct SC at baseline and follow-up in patients receiving radiation therapy for brain metastases and bone metastases.
In a study of 1296 advanced cancer patients, Chen et al. also found depression and anxiety as a distinct category of symptom distress, and they supported their data using three statistical methods (principal component analysis, hierarchical cluster analysis, and exploratory factor analysis)(Chen et al., Reference Chen, Nguyen and Cramarossa2012). In a study of 1366 advanced cancer patients, Cheung et al. identified two SCs at an outpatient palliative care clinic. Similar to us, they found a physical cluster that included fatigue, drowsiness, nausea, decreased appetite, and dyspnea and a psychological cluster that included anxiety and depression (Cheung et al., Reference Cheung, Le and Zimmermann2009). In that study, anxiety and depression continued to constitute a cluster even when patients with solid tumors were partitioned into subgroups on the basis of the cancer site.
In another study of 1296 patients with advanced cancer who were seen at an outpatient palliative radiation clinic, Fan et al. found two distinct physical SCs (cluster 1 included lack of appetite, nausea, poor sense of well-being, and pain; cluster 2 included fatigue, drowsiness, and shortness of breath) and a psychological cluster (depression and anxiety) (Fan et al., Reference Fan, Hadi and Chow2007).The authors of that study reported that anxiety and depression had no relationship to the physical symptoms.
Even though the patient populations were different, all studies cited found that depression and anxiety combined to form a distinct SC. These results suggest that all advanced cancer patients receiving outpatient palliative care require supportive counseling in addition to physical symptom management. Future studies using multimodal strategies to determine the optimal use of an interdisciplinary team should be considered.
Prior studies reported interventions using SCs as an outcome measure. The responses to these interventions varied according to the type of intervention (Jarden et al., Reference Jarden, Nelausen and Hovgaard2009; Dodd et al., Reference Dodd, Cho and Miaskowski2010; Husain et al., Reference Husain, Bezjak and Easson2010; Kwekkeboom et al., Reference Kwekkeboom, Abbott-Anderson and Wanta2010; Chan et al., Reference Chan, Richardson and Richardson2011) Improvement of SCs after intervention was reported in the trials with multidimensional management (Jarden et al., Reference Jarden, Nelausen and Hovgaard2009) or a pathophysiology-based approach (Husain et al., Reference Husain, Bezjak and Easson2010; Chan et al., Reference Chan, Richardson and Richardson2011), whereas a single modality approach did not cause improvement in SCs (Dodd et al., Reference Dodd, Cho and Miaskowski2010; Kwekkeboom et al., Reference Kwekkeboom, Abbott-Anderson and Wanta2010). These data suggest that either a multimodal- or a pathophysiology-based approach to SCs could be reasonable in management of SCs (Cheville et al., Reference Cheville, Novotny and Sloan2011). Prior studies have used biological markers (Steel et al., Reference Steel, Kim and Dew2010) and the number of SCs (Jiménez et al., Reference Jiménez, Madero and Alonso2011) have been suggested as indicators of clinical outcomes in the improvement of quality of life. Our study did not analyze the SC as an indicator of outcomes. However, the consistency during follow-up of the order of severity of each symptom and the stability of the SCs suggest that SCs could be used as effective outcome measures. Further prospective studies are needed to validate these findings.
Limitations
Our study had many limitations. Our study used the 10 item ESAS score to identify symptom candidates for the cluster analysis. Considering that a median of 11 symptoms was reported in cancer patients (Walsh & Rybicki, Reference Walsh and Rybicki2006) other symptoms also may have shown high correlation with symptoms in the ESAS, including altered sexuality (Wilmoth et al., Reference Wilmoth, Coleman and Smith2004), cognitive impairment (Fox et al., Reference Fox, Lyon and Farace2007), weight loss (Francoeur, Reference Francoeur2005), or some side effects of chemotherapy such as mucositis. Also, although our study was based on prospectively collected ESAS scores, the analysis was performed retrospectively. The absence of a control group in our study could be another limitation; however, the data presented have significance, as the patients with advanced cancer showed improvement rather than worsening of symptoms upon follow-up, in contrast to what is commonly expected in advanced cancer as a result of disease progression. The higher baseline mean intensity of the physical SC score compared with the psychological SC score (p < 0.0001) in our study suggests that the physical distress of our patient population was more severe than the psychological distress; however, it also is possible that we were unable to capture the entire dimension and severity of psychological distress using ESAS anxiety and depression items as compared with validated tools such as the Hospital Anxiety and Depression Scale.
Because of the wide use of ESAS among palliative care clinicians, an effort to find SCs in the ESAS and to develop care plans addressing ESAS clusters will have great implications in the clinical world. For example, there is a paucity of studies about managing symptoms of cancer patients according to SCs.
Several biologic markers have been suggested in the pathophysiology of cancer-related symptoms. The C-reactive protein as a marker for systemic inflammation is known to be associated with cancer-related pain (Rokyta et al., Reference Rokyta, Haklova and Yamamotova2009; Laird et al., Reference Laird, Scott and Colvin2011a), SCs (e.g., pain, depression, and fatigue) (Laird et al., Reference Laird, Scott and Colvin2011b), and anxiety and depression (Brown et al., Reference Brown, McMillan and Milroy2005). Other markers including interleukins such as Interleukin-6 also have been associated with cancer pain (Starkweather et al., Reference Starkweather, Lyon and Schubert2011). The antitumor necrosis factor alpha (anti-TNF-α) antibody also has been suggested as a biologic marker of the pathophysiology of cancer-related symptoms (Beutler & Cerami, Reference Beutler and Cerami1988; Argilés & López-Soriano, Reference Argilés and López-Soriano1999; Kurzrock, Reference Kurzrock2001) For example, an anti-TNF-α antibody was tested in cancer-related fatigue and conferred marginal benefit, which may lead one to assume that some SCs have a common pathophysiology; future studies to characterize the pathophysiology of SCs in advanced cancer and to manage the patient's symptoms according to the SCs (e.g., use of anti-inflammatory agents such as dexamethasone or thalidomide for management of physical SCs and psychotherapy and anti-inflammatory agents for psychological SCs) are needed.
CONCLUSION
In conclusion, we identified two SCs, physical and psychological, at the initial visit, and found that these SCs were consistent at the first follow-up visit. The intensity of these physical and psychological SCs decreased, however. This knowledge may allow palliative care teams to provide more targeted and higher-quality care, but further studies are needed.
ACKNOWLEDGMENT
We thank Ms. Luanne Jorewicz for manuscript review.
