A Patient‐Reported Outcome Instrument to Assess Symptom Burden and Predict Survival in Patients with Advanced Cancer: Flipping the Paradigm to Improve Timing of Palliative and End‐of‐Life Discussions and Reduce Unwanted Health Care Costs

Discussions regarding palliative care and end‐of‐life care issues are often delayed past the time of usefulness, resulting in unwanted medical care. This article reports the results of a multicenter study with a patient‐reported outcome tool used with more than 1,000 patients with advanced cancers to assess the association between patient symptom burdens distress and overall survival.

Abstract

Background.

Discussions regarding palliative care and end‐of‐life care issues are frequently delayed past the time of usefulness, resulting in unwanted medical care. We sought to develop a patient‐reported outcome (PRO) instrument that allows patients to voice their symptom burdens and facilitate timing of discussions.

Subjects, Materials, and Methods.

A seven‐item PRO instrument (Cota Patient Assessed Symptom Score‐7 item [CPASS‐7]) covering physical performance status, pain, burden, and depression was administered (September 2015 through October 2016) with correlation to overall survival, correcting for time to complete survey since diagnosis.

Results.

A total of 1,191 patients completed CPASS‐7 at a median of 560 days following the diagnosis of advanced cancer. Of these patients, 49% were concerned that they could not do the things they wanted; 35% reported decreased performance status. Financial toxicity was reported by 39% of patients, with family burdens noted in 25%. Although depression was reported by 15%, 43% reported lack of pleasure. Pain was reported by 33%. The median CPASS‐7 total symptom burden score was 16 (possible 0–112). With a median follow‐up of 15 months from initial survey, 46% had died. Patients with symptom burden scores <29 and ≥29 had a 6‐month overall survival rate of 87% and 67%, respectively, and 12‐month survival rates of 72% and 50%. A one‐point score increase resulted in a 1.8% increase in expected hazard.

Conclusion.

Patients with advanced cancer with higher levels of symptom burden, as self‐reported on the CPASS‐7, had inferior survival. The PRO facilitates identification of patients appropriate for reassessment of treatment goals and potentially palliative and end‐of‐life care in response to symptom burden concerns.

Implications for Practice.

A seven‐item patient‐reported outcome (PRO) instrument was administered to 1,191 patients with advanced cancers. Patients self‐reporting higher levels of physical and psychological symptom burden had inferior overall survival rates. High individual item symptom PRO responses should serve as a useful trigger to initiate supportive interventions, but when scores indicate global problems, discussions regarding end‐of‐life care might be appropriate.

Introduction

The Institute of Medicine's Committee on Care at the End of Life defines a “good death” as one “free from avoidable distress and suffering for patients, families, and caregivers; in general accord with the patients' and families' wishes; and reasonably consistent with clinical, cultural, and ethical standards.” [1]. To achieve this goal and honor the wishes of the terminally ill patient, discussions regarding advance care planning, palliative care, and end‐of‐life care (EOLC) must be undertaken at the appropriate time. However, frequently physicians are reluctant to initiate these discussions and/or may provide unrealistic prognostic expectations [2], [3], [4], [5], [6], [7], [8].

Patients may also be hesitant to discuss EOLC issues with their medical team [9], [10], [11], [12]. In his book Being Mortal, Atul Gawande writes that “We imagine that we can wait until the doctors tell us that there is nothing more they can do. But rarely is there nothing more that doctors can do.” [13]. Reviews of cancer patient admission documents demonstrate that less than 10% have discussed their EOLC plans with their clinic oncologist and less than a quarter want to discuss their EOLC plans with their oncologists [12], [14]. As a consequence, discussions about EOLC are frequently delayed past the point of usefulness, resulting in expensive unwanted medical care, as well as reduced quality of life for patients and bereavement adjustment for caregivers [15], [16]. This is occurring despite approximately half of cancer patients having advanced directives completed and over two thirds having discussed their wishes with family members.

The lack of timely EOLC discussions may also have a significant economic impact [17]. The cost of health care rises steeply as end of life approaches, with nearly one third of Medicare expenditures during the last year of life occurring within the last month of life [18], [19]. Among cancer patients, hospital services are the main cost drivers, accounting for over half the total expenditures during the last 6 months of life and rising nearly 10‐fold during the last month [20], [21]. Chemotherapy and other life‐sustaining treatments are administered to 9%–38% of patients in the last month of life. And over half of cancer patients who die do not receive hospice or palliative care services [22], [23]. By contrast, patients who have palliative and EOLC discussions with their physicians are less likely to experience symptom burdens at life's end, less likely to undergo high‐intensity interventions, less likely to die in the intensive care unit, and more likely to receive outpatient hospice care and be referred to hospice earlier [24], [25]. Overall, patients who communicate their preferences to their physicians are more likely to receive care consistent with their desires [26], [27].

Prior attempts to encourage advance care planning among cancer patients have focused on improving the accuracy of identifying patients at risk of imminent death [28], [29], [30], [31]. Armed with survival projections, the health care team may be more likely to approach patients to initiate EOLC discussions. Additionally, economic incentives to encourage these palliative and EOLC discussions have been introduced [32]. Medicare's Merit‐based Incentive Payment System penalizes physicians for failure to appropriately use hospice services, for excessive use of intensive care, or emergency care services within the final 30 days of life, or for administration of chemotherapy within the final 2 weeks of life [33]. Recently, the American Society of Clinical Oncology (ASCO) revised its guidelines to encourage early palliative care consultation within 8 weeks of diagnosis and throughout active cancer treatment [34]. Universal early palliative care discussions, however, do run the risk of confusing patients who are facing complex treatment decisions.

What is missing from these efforts is the patient's voice. Moving from delayed, last‐minute EOLC conversations to ASCO's universal early palliative discussions is based on health care professional input, not patient wishes. Patient‐reported outcome (PRO)‐guided continuous monitoring of patient symptom burden has been associated with improved survival among patients with advanced cancer [35]. The Cota Patient Assessed Symptom Score‐7 item (CPASS‐7) PRO instrument was designed to elicit levels of common causes of physical and psychological symptom burden among patients with advanced diseases, without specifically mentioning palliative care or EOLC. The goal was to develop an instrument that “flips the paradigm” for timing of palliative and EOLC discussions, not by encouraging physicians to raise the subject based on physical signs of poor projected outcomes, but rather by asking patients about symptom burdens, thereby giving the patient the opportunity to express their concerns. Complaints regarding a single symptom burden might be best addressed with appropriate symptom interventions. However, we hypothesize that patients reporting global burdens might actually be expressing concerns that the current medical strategy is not aligned with their goals and that discussions regarding alternative treatment strategies might be more appropriate. We hoped that this instrument would liberate physicians from the subjectivity around timing of palliative and EOLC discussions and better equip them for these conversations. A pilot study with the 7‐item instrument among 433 patients with cancer at a single institution demonstrated that higher levels of symptom burdens independently correlated with physician assessments for the appropriateness of EOLC discussion [36]. We report here the results of a multicenter study with the PRO tool among over a thousand patients with advanced cancers to assess the association between patient symptom burdens and overall survival.

Subjects, Materials, and Methods

Patient‐Reported Outcome Instrument

A seven‐item PRO study instrument, CPASS‐7, covers four personhood and clinical domains: physical performance status [37], [38], [39], pain [40], burden (financial and family) [41], [42], and depression [43], [44], [45] (Fig. 1). Each of the instrument's seven items is measured on a five‐level Likert scale scored 0 to 4. A weighting based on the patient's perception of the importance of the item (scored 1, 2, or 4 for “not at all,” “moderately,” or “significantly” important) is recorded; the total CPASS‐7 weighted score ranges from 0 to 112 (maximum score of 4 on each item times 4 for importance times 7 items). Surveys that were returned with three or more of the seven items not completed were discarded. Surveys with two or fewer items left uncompleted were included and the missing items were given a score of zero (An exploratory review using median value imputation last value carried forward did not significantly alter the results).

Figure 1.

The Cota Patient Assessed Symptom Score‐7 item Instrument.

Prior Validation Phases of the CPASS‐7

A pilot study on 433 patients with advanced cancer identified a potential threshold score of 29 discriminating patients for whom their oncologists felt end‐of‐life care discussions were appropriate [36]. A statistical validation was performed using responses from 700 patients with advanced cancer (partially overlapping with this study's cohort), using the oncologist's blinded opinion of the patient's outlook as the reference standard, and confirmed an optimal threshold score of 29, with sensitivity of 51%, specificity of 78%, positive predictive value of 25%, and false omission rate of 8%. CPASS‐7 had a Cronbach's alpha coefficient of 0.85, indicating internal consistency and ability to measure a single latent construct [46].

Patient Population

The CPASS‐7 PRO Instrument was offered to patients with advanced malignancies (metastatic or systemic disease) receiving noncurative intent treatment at seven outpatient cancer centers within New Jersey's Regional Cancer Care Associates network from September 2015 to October 2016. The PRO was administered in the waiting rooms or clinic rooms without additional patient instructions. No attempts to control for physician discretion regarding not offering surveys to particular patients or to confirm the total metastatic cancer denominator population were made. The physicians and medical team were able to review the results with the patient immediately upon completion so that symptom interventional support could be offered. CPASS‐7 questionnaires with more than three items left blank were not analyzed. During the study timeframe, patients may have completed multiple CPASS‐7 PROs; however, in this analysis, only the first qualifying PRO results are reported. A “time ratio” normalizing the time from diagnosis to survey completion divided by the time from diagnosis to date of death (or last follow‐up censor) was calculated, with a ratio of 0 equaling survey completed at date of diagnosis and 1.0 equaling survey completed on date of death/censor. This calculation is meant to help define where the patient was within the expected course of the disease (newly diagnosed, middle of disease, nearing death, etc.).

Consent

The concept and study design were approved by the Human Institutional Review Board at Hackensack University Medical Center and the study was deemed to be exempt from the need for informed consent. The seven items addressed on the PRO were already being collected within the provider organizations as part of routine oncologic care. The physicians and medical team were able to immediately view the results of the PRO at the time of collection so that appropriate supportive interventions could be taken to reduce symptom burdens. The results of the PRO along with the entire electronic record were reviewed, under Business Associate Agreements, by Cota for the provider groups to support additional quality and insurance/financial programs including participation in the Oncology Care Model and several commercial insurer value‐based payment programs. The results of the PRO and the abstracted medical record were later deidentified for secondary research purposes before being given to the study team for analysis. The Cota Observational Database, which tracks patients longitudinally through their care (including survival data), has been reviewed by the Western institutional review board.

Statistical Analysis

Variables were summarized using frequencies and proportions for categorical variables, and means, medians, and interquartile ranges for continuous variables. Time to event analysis was done using date of survey as the index date and date of death or last follow up as the endpoint. The Kaplan–Meier method was used to estimate survival probabilities, and the resulting curves were compared using the log‐rank test. Cox proportional hazards models were used to assess the association of predictors with overall survival. Proportionality assumption was reviewed by testing for a nonzero slope in a generalized linear regression of the scaled Schoenfeld residuals on functions of time. The results demonstrated that the proportionality assumption was not violated. All analysis was done using R 3.3.2.

Results

Patients Demographics

The CPASS‐7 instrument was completed by 1,191 patients with advanced cancers. No information is available on the number of patients who declined to complete the instrument or patients the physicians chose not to offer the survey. Less than 10% of surveys were discarded for >3 missing items and no replacement survey available at the next visit. Three patients who failed to complete an initial survey (≥3 items missing) subsequently fully completed a replacement survey. Patient demographics are shown in Table 1. The most common malignancies were gastrointestinal (n = 331, 28%), multiple myeloma (n = 239, 20%), breast cancer (n = 171, 14%), thoracic (n = 143, 12%), and genitourinary (n = 140, 12%).

Table 1. Patient demographics.

Abbreviations: CPASS‐7, Cota Patient Assessed Symptom Score‐7 item; IQR, interquartile range.

Timing of PRO

The CPASS‐7 PRO was completed a median of 560 days (18.7 months) following the diagnosis of advanced disease and differed by cancer subtype. Only the first completed (nearer to diagnosis) responses were used for this analysis. Patients with myeloma completed the PRO a median of 1,976 days following diagnosis, with breast 698 days, genitourinary 585 days, gastrointestinal 303 days, and thoracic 257 days following a metastatic diagnosis. Among patients who have died, the PRO was completed a median of 423 days prior to death. The PRO was completed at a median time ratio of 0.63 (i.e., on average 63% of the median life expectancy for their cancer subtype), with myeloma patients completing the survey later in the course of their disease (time ratio 0.81), followed by genitourinary 0.64, breast 0.63, gastrointestinal 0.53, and thoracic tumors 0.47 time ratio, respectively.

Symptom and Psychologic Concerns as Self‐Reported on the CPASS‐7 Instrument

Forty‐nine percent of patients were concerned that they could not do the things they wanted to do (ratings ≥2), and 35% reported decreased performance status. Financial toxicity concerns were self‐reported by 39%, with family burdens in 25%. Although depression was reported by only 15%, 43% reported lack of pleasure. Pain was self‐reported by 33%. Similar patterns of concern were noted among diseases, although patients with thoracic cancer had more symptom burdens (p < .01; Fig. 2).

Figure 2.

Likert level scores for each question on the Cota Patient Assessed Symptom Score‐7 item by major malignancies. Patients rating an item as moderate, significant, or severe were considered to be concerned about this item. Abbreviation: ECOG PS, Eastern Cooperative Oncology Group Performance Status.

Overall Survival Associations with CPASS‐7 Total Symptom Burden Scores

The median CPASS‐7 total symptom burden score, combining the seven Likert items and their associated patient weightings, was 16 (interquartile range 7‐30). In the prior pilot study, a threshold score ≥29 was found to be associated with the physician's opinion regarding appropriateness to initiate end‐of‐life discussions. Three hundred twelve patients (26%) exceeded this threshold (Fig. 3).

Figure 3.

Box plot of patient total distress scores calculated by adding the sums of each item's response times the question's weighting (possible range 0–112).

With a median follow‐up of 15 months from initial survey (study cutoff date August 8, 2017), 549 patients (46%) had died, with a median time to death of 8 months following completion of the first CPASS‐7 survey (breast cancer 33% expired median follow‐up 15.9 months; gastrointestinal cancer 56% expired at 14.1 months; genitourinary 43% expired at 11.6 months; myeloma 28% expired at 15.9 months; thoracic cancers 69% at 13.2 months).

Patients with a CPASS‐7 total symptom burden score <29 had a 6‐month overall survival rate of 87% and a 12‐month overall survival rate of 72%, compared with patients with a score ≥29, who had 6‐ and 12‐month overall survival rates of 67% and 50%, respectively (log‐rank p < .001; Fig. 4). The threshold value of 29 was discriminatory for patients with gastrointestinal and genitourinary malignancies at p < .001, and for breast, myeloma, and thoracic oncology at p < .1. The 6‐month survival for patients with CPASS‐7 total symptom burden scores <29 and ≥29 with breast cancer was 93% and 84%, with gastrointestinal cancers was 94% and 57%, with genitourinary cancers was 87% and 66%, with myeloma was 96% and 88%, and with thoracic tumors was 74% and 52%, respectively. A Cox model on the top five cancers demonstrated that the CPASS‐7 score and cancer type were significant overall (CPASS‐7: p < .001, cancer types (compared with breast): gastrointestinal p < .001, genitourinary p < .01, thoracic p < .001, myeloma p = .12). A one‐point score increase in CPASS‐7 resulted in a 1.8% increase in expected hazard. Among patients with solid tumors with CPASS‐7 ≥ 29, 10% died within the next 3 months and 22% died within the next 6 months.

Figure 4.

Kaplan‐Meyer overall survival estimates of patients with advanced cancers stratified by total distress scores as self‐reported on the Cota Patient Assessed Symptom Score‐7 item.

Discussion

Traditional approaches to addressing end‐of‐life concerns have focused on the identification of high‐risk disease features that signal impending death, thereby enabling health care personnel to identify patients unlikely to benefit from additional aggressive treatment. This strategy, however, may place the health care team in an uncomfortable position, seen as limiting care or removing hope. By contrast, the seven‐item, patient‐centric CPASS‐7 instrument was designed to flip the paradigm and encourage patients with advanced cancer to express common issues of symptom burden. Armed with this information, the health care team hopefully will be able to address individual items of concern or, when total symptom burden is high, offer to reassess treatment goals and palliative care issues and potentially initiate end‐of‐life support. In the current multi‐institutional study of 1,191 patients with advanced cancer, the CPASS‐7 instrument was able to establish a threshold total symptom burden score, which was associated with inferior survival, adding to the information needed by health care personnel in facilitating palliative care or end‐of‐life discussions. The threshold survival score in the current study was similar to a prior pilot project in which physicians blinded to the CPASS‐7 results identified patients with cancer for whom they thought end‐of‐life conversations would be appropriate, suggesting that physicians may be able to identify patients with inferior survival in need of EOLC consultations [36]. The CPASS‐7 was most predictive in solid tumors, diseases in which symptom burdens increase with disease progression, compared with myeloma. We suspect that diseases in which symptoms do not strongly track with disease progression may not be optimal for this approach.

The study had several limitations that need to be considered. We did not control administration of the CPASS‐7 instrument at the multiple cancer centers. It is possible that physicians may have been selective in offering the survey to patients and we do not have data on the number of patients who refused to complete the instrument. We suspect that the most ill patients may not have completed the survey (or even been offered the instrument), potentially underestimating the incidence and severity of symptom burden in the total cancer population. We also do not know how symptom burden scores change over time during repeated surveys, a potentially important observation during the continuum of care. Although we envision the instrument being of assistance to physicians in starting palliative or EOLC discussions, the study did not examine the outcomes of such interventions.

The CPASS‐7 instrument assesses four personhood and clinical domains by the patient with advanced cancer, without specifically describing end of life. The instrument is able to alert the health care team at a granular level of correctable concerns and to be used on a repeated nature basis to assess changes in concerns over time. The patterns of concern identified in this study were relatively consistent across tumor types. The inability to do the things that they wanted to do was cited as a symptom burden in almost half of all respondents, exceeding performance status, which has been one of the strongest predictors of survival among cancer patients [37], [38], [39], [40]. Financial toxicity also emerged as an important stressor in cancer care and was cited by over a third of patients. The ability to repeat the PRO instrument and assess financial issues (and family burdens) over time may also be critical as costs and time off work accrue. A prior review with this PRO in patients with cancer noted that younger individuals and those living in less affluent neighborhoods were more concerned about the costs of their care [47].

The total CPASS‐7 symptom burden score was associated with survival in this cohort of patients at 6 and 12 months. Previously, using a cohort of 1,302 patients with advanced cancer (some overlapping with this study), we performed a latent factor analysis and noted that five items in the PRO (excluding financial burden and depression) were also predictive of 3‐month survival (p < .001) [48].

As a potential quality indicator or metric, the CPASS‐7 might be used to benchmark physician palliative care referral patterns. The quantitative nature of the instrument has the potential to facilitate timely consultations triggered by the survival threshold score. Responses to changes in the total symptom burden score, which is within the control of the health care team, can be objectively measured, as opposed to the current system of rating physicians by hospice days or treatments within 30 days of death (which may be outside physician control). By gaining an understanding of the patient's current levels of symptom burden, the timing of palliative care consultation can be tailored to patient needs. This personalized approach might be viewed as at odds with recent trends encouraging routine application of early palliative care consultations for patients with serious diseases [34], [49], [50]. We are not averse to early palliative care support but believe the CPASS‐7 allows the referrals to address true areas of concern at the appropriate time. A recent meta‐analysis of 43 randomized controlled trials that provided information on 12,731 patients demonstrated that palliative care was associated with statistically and clinically significant improvements in quality‐of‐life parameters and symptom burden. Timely palliative care was consistently associated with improvements in advance care planning, patient and caregiver satisfaction, and lower health care use, but no association between palliative care and survival was observed [51]. The PRO instrument, which can be used to repeatedly assess multiple patient domains, supports the recent ASCO guidelines that encourage physicians to discuss patient concerns, reassess goals and costs of therapy, and address EOLC issues continually through the course of care [52]. However, as noted above, we did not assess the value of repetitive administration of the PRO over time.

As medicine continues to move more toward value‐based payment reform, the cost of care at end of life will be increasingly scrutinized. A meta‐analysis of 78 studies involving over 3.7 million patients with cancer found exponential increases in service use and costs as death approached, hospital services being the main cost driver. Palliative care services were underused and associated with lower expenditures than hospital‐based care. In the 15 studies using quality indicators, up to 38% of patients received chemotherapy or life‐sustaining treatments in the last month of life and up to 66% did not receive hospice/palliative services [22]. Timely discussions regarding EOLC might curtail unwanted expensive cancer care [53]. A longitudinal multi‐institutional cohort study, Coping With Cancer, demonstrated that these discussions led to fewer life‐sustaining procedures, lower rates of intensive care unit admission, and substantially lower total costs [24], [54]. The CPASS‐7 instrument, by encouraging appropriately timed palliative and EOLC discussions as patient‐reported global symptom burden worsens, may curtail expensive terminal hospitalizations.

Conclusion

The C‐PASS‐7 PRO allows for continuous monitoring of symptom burdens among patients with advanced cancers. Although identification of a single‐item burden on the PRO may be potentially best addressed with symptom‐directed interventions, a high total symptom burden score was associated with inferior survival and should trigger a discussion regarding alignment of treatment goals, palliative care issues, and potentially EOLC. Most importantly, because the CPASS‐7 expresses the patient's voice, the health care team will not be viewed as denying care but rather responding to the patient's symptom burden and aligning care with patient wishes—a primary duty of health care professionals.

Author Contributions

Conception/design: Stuart L. Goldberg, Dhakshila Paramanathan, Raya Khoury, Sharmi Patel, Dayo Jagun, Srikesh Arunajadai, Victoria DeVincenzo, Ruth Pe Benito, Brooke Gruman, Sukhi Kaur, Scott Paddock, Andrew D. Norden, Eric V. Schultz, John Hervey

Provision of study material or patients: Stuart L. Goldberg, Terrill Jordan, Andre Goy, Andrew L. Pecora

Collection and/or assembly of data: Stuart L. Goldberg, Dhakshila Paramanathan, Srikesh Arunajadai, Victoria DeVincenzo, Ruth Pe Benito, Brooke Gruman, Sukhi Kaur, Scott Paddock, Andrew D. Norden, Eric V. Schultz, John Hervey

Data analysis and interpretation: Stuart L. Goldberg, Dhakshila Paramanathan, Raya Khoury, Sharmi Patel, Dayo Jagun, Srikesh Arunajadai, Victoria DeVincenzo, Ruth Pe Benito, Brooke Gruman, Sukhi Kaur, Scott Paddock, Andrew D. Norden, Eric V. Schultz, John Hervey, Terrill Jordan, Andre Goy, Andrew L. Pecora

Manuscript writing: Stuart L. Goldberg, Dhakshila Paramanathan, Raya Khoury, Sharmi Patel, Dayo Jagun, Srikesh Arunajadai, Victoria DeVincenzo, Ruth Pe Benito, Brooke Gruman, Sukhi Kaur, Scott Paddock, Andrew D. Norden, Eric V. Schultz, John Hervey, Terrill Jordan, Andre Goy, Andrew L. Pecora

Final approval of manuscript: Stuart L. Goldberg, Dhakshila Paramanathan, Raya Khoury, Sharmi Patel, Dayo Jagun, Srikesh Arunajadai, Victoria DeVincenzo, Ruth Pe Benito, Brooke Gruman, Sukhi Kaur, Scott Paddock, Andrew D. Norden, Eric V. Schultz, John Hervey, Terrill Jordan, Andre Goy, Andrew L. Pecora

Disclosures

Stuart L. Goldberg: Cota Inc (E), Genentech (H); Dhakshila Paramanathan: Cota Inc (E); Raya Khoury: Genentech (E); Sharmi Patel: Genentech (E); Dayo Jagun: Genentech (E); Srikesh Arunajadai: Cota Inc (E); Victoria DeVincenzo: Cota Inc (E); Ruth Pe Benito: Cota Inc (E); Brooke Gruman: Cota Inc (E); Sukhi Kaur: Cota Inc (E); Scott Paddock: Cota Inc (E); Andrew D. Norden: Cota Inc (E); Eric V. Schultz: Cota Inc (E); John Hervey: Cota Inc (E); Andrew L. Pecora: Cota Inc (OI). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board