Effects of an Oncology Nurse-Led, Primary Palliative Care Intervention (CONNECT) on Illness Expectations Among Patients With Advanced Cancer

PURPOSE:

Patients with advanced cancer often have unrealistic expectations about prognosis and treatment. This study assessed the effect of an oncology nurse-led primary palliative care intervention on illness expectations among patients with advanced cancer.

METHODS:

This study is a secondary analysis of a cluster-randomized trial of primary palliative care conducted at 17 oncology clinics. Adult patients with advanced solid tumors for whom the oncologist would not be surprised if died within 1 year were enrolled. Monthly visits were designed to foster realistic illness expectations by eliciting patient concerns and goals for their medical care and empowering patients and families to engage in discussions with oncologists about treatment options and preferences. Baseline and 3-month questionnaires included questions about life expectancy, treatment intent, and terminal illness acknowledgment. Odds of realistic illness expectations at 3 months were adjusted for baseline responses, patient demographic and clinical characteristics, and intervention dose.

RESULTS:

Among 457 primarily White patients, there was little difference in realistic illness expectations at 3 months between intervention and standard care groups: 12.8% v 11.4% for life expectancy (adjusted odds ratio [aOR] = 1.15; 95% CI, 0.59 to 2.22; P = .684); 24.6% v 33.3% for treatment intent (aOR = 0.76; 95% CI, 0.44 to 1.27; P = .290); 53.6% v 44.7% for terminal illness acknowledgment (aOR = 1.28; 95% CI, 0.81 to 2.00; P = .288). Results did not differ when accounting for variation in clinic sites or intervention dose.

CONCLUSION:

Illness expectations are difficult to change among patients with advanced cancer. Additional work is needed to identify approaches within oncology practices that foster realistic illness expectations to improve patient decision making.

INTRODUCTION

Illness expectations refer to a patient's perception of his or her prognosis, treatment aims, and illness trajectory,^1,2 which are often unrealistic among patients with advanced cancer. In one study, 69%-81% of patients with metastatic lung and colorectal cancers reported that their cancer might be cured by chemotherapy.^3,4 Unrealistic illness expectations may be problematic because patients make different medical decisions on the basis of their expectations of prognosis, treatment aims, and illness trajectory.^3,5-8 More realistic patient illness expectations are associated with transitions in medical care that better reflect patient preferences, earlier hospice referral, and decreased health care resource utilization near the end of life.^1,4,5,7,9-11

Specialty palliative care promotes discussions about patient prognostic awareness, goals, and future planning and has been associated with improved patient coping abilities and communication regarding end-of-life preferences when provided alongside standard oncology care for patients with advanced cancer.^10,12,13 For instance, Temel et al^12,13 demonstrated that for patients with metastatic non–small-cell lung cancer, early specialty palliative care in addition to standard oncology care significantly improved the accuracy of patient perceptions of curability when compared with standard oncology care alone (82.5% v 59.6%; P = .02). Those patients who understood and acknowledged their overall poor prognosis had significantly less chance of receiving intravenous chemotherapy in their last 2 months of life (9.4% v 50%; P = .02). However, this study did not demonstrate a difference between groups with respect to patient perceptions of goals of therapy, with a majority of patients in both groups continuing to report that a goal of therapy was to get rid of all the cancer.¹²

ASCO recommends integrating palliative care for all critically ill oncology patients at the time of diagnosis.¹⁵ However, limited availability of specialty palliative care and inconsistent referral practices often prevent patients with advanced cancer from receiving specialty palliative care.^9,12,14-18 Primary palliative care, which is provided by clinicians without subspecialty palliative care training or board certification,¹⁹ provides an opportunity to expand access to palliative care resources for patients with advanced cancer.^20-27 Primary palliative care incorporates fundamental tenets of palliative care, including symptom management, advance care planning, and care coordination, and is often provided by nurses in either inpatient or home settings.^19,24,28 Evidence from a recent systematic review suggests that primary palliative care approaches are associated with improved patient quality of life and mood without adverse survival outcomes.¹⁹ However, to date, studies of primary palliative care interventions in oncology have not assessed impacts on patient illness expectations.^29,30

This analysis investigates the role of primary palliative care on illness expectations in patients with advanced cancer. Care Management by Oncology Nurses to address supportive care needs (CONNECT) is a primary palliative care intervention led by existing oncology nurses that focuses on symptom assessment and management, emotional support, advance care planning, and care coordination in patients with advanced cancer.^31-33 CONNECT nurses elicit patient concerns and goals for their medical care and empower patients and families to engage in discussions with oncologists about treatment options and preferences. We hypothesized that CONNECT might lead to more realistic illness expectations in patients with advanced cancer by (1) assessing patient illness views and providing opportunities for patients and families to reflect on these views and (2) supporting patients and families in discussing questions with their oncologists via shared care plans and oncologist check-ins.

METHODS

Study Design

This analysis uses data from the CONNECT cluster-randomized trial (NCT02712229), in which medical oncology practices were randomized to intervention or standard care groups.^31,32,34 CONNECT sought to investigate the impact of incorporating primary palliative care with standard oncology care on patient, caregiver, and health care outcomes for patients with incurable cancer. Details including the study design and intervention feasibility were previously published.^31,32 However, to our knowledge, no prior CONNECT analysis has assessed the intervention's impact on illness expectations. Approval was obtained from the University of Pittsburgh Institutional Review Board (PRO15120154).

Participants

Patients were recruited from 17 participating medical oncology clinics within the UPMC Hillman Cancer Center Network in western Pennsylvania from July 2016 through October 2019. Patients were eligible if they met the following inclusion criteria: adults age ≥ 21 years, metastatic solid tumor diagnosis, primary outpatient oncologist would not be surprised if patient died in the next year,^35-37 Eastern Cooperative Oncology Group (ECOG) status ≤ 2 (0 = fully active, 1 = restricted strenuous activity but ambulatory and capable of light or sedentary work, and 2 = ambulatory and capable of self-care),³⁸ planning to receive ongoing care from an oncologist at a participating site, and agreeable to monthly intervention visits. Patients were excluded if they were unable to read and respond to questions in English, had cognitive impairment or inability to consent to treatment as determined by their oncologist, were unable to complete the baseline interview, had ECOG performance status of 3 (capable of only limited self-care; confined to bed or chair > 50% of waking hours) or 4 (cannot carry on any self-care; totally confined to bed or chair),³⁸ or had hematologic malignancies. Only patients with baseline and 3-month data assessments were included in the final analysis; extrapolating responses for patients who were deceased or lost to follow-up would likely introduce biases because of the impossibility of predicting their expectations over time.

The sample size for the CONNECT trial was determined on the basis of the primary outcome measure (quality of life).³⁹ Here we present power calculations for the illness expectation measures, which were not carried out before study design but are made available to contextualize the findings presented in this article. In the standard care group, the proportion of participants giving the correct response to the illness expectation questions at 3 months were 11.4% for life expectancy, 35.8% for treatment intent, and 44.7% for terminal illness acknowledgment. With an available sample size of 211 CONNECT participants and 246 standard care participants who completed 3-month assessments with complete illness expectation data, this secondary analysis would have at least 80% power to detect treatment effects if the treatment group's true probabilities of correct response were 21% for life expectancy (up from 11.4% in the standard care group), 49% for treatment intent (up from 35.8% in the standard care group), and 58% for terminal illness acknowledgment (up from 44.7% in the standard care group).

CONNECT Intervention

CONNECT is an oncology nurse-led, primary palliative care and care management intervention designed to leverage existing relationships between oncology nurses, patients, and oncologists.^32,40,41 Details on nurse selection and training were previously published.^31,32 The intervention consisted of nurse-driven encounters that occurred before or after patients' regularly scheduled outpatient oncology appointments, either in person or via phone, with visits planned monthly for 3 months. Each visit included standard interaction checklists that were completed by nurses and monitored by trained raters for consistency. The initial visit was aimed at building rapport, addressing symptom needs, and identifying surrogate decision makers. Subsequent encounters focused on patient views on illness, ongoing symptom management, and advance care planning, including identifying treatment preferences and goals and completing a shared care plan. CONNECT nurses did not disclose prognosis or attempt to correct illness expectations but rather elicited patients' concerns and goals, promoted coping strategies, and empowered patients and families to engage in discussions with oncologists about treatment options and preferences. After each visit, nurses communicated directly with treating oncologists to facilitate individualized patient-oncologist discussions about patient concerns. Of the 211 CONNECT participants included in this analysis, 119 patients completed all three planned intervention visits before the 3-month assessment, 46 completed two intervention visits, 24 completed one intervention visit, and 22 completed no intervention visits before the 3-month assessment.

Illness Expectations Assessment

Illness expectation questions were prespecified secondary outcomes of the CONNECT trial.³⁹ Illness expectations were evaluated using questions adapted from prior palliative care studies in patients with advanced cancer.^9,13,42 Baseline and 3-month questionnaires assessed patient perceptions of life expectancy, treatment intent, and terminal illness acknowledgment.^9,11,13 Patients were asked the following questions with responses defined as realistic indicated in italics.

1. Life expectancy was assessed using the question “When you think about how having cancer might affect your life expectancy, do you think in terms of: months, years, or do not know,” adapted from Coping with Cancer and the VOICE trial.^9,43

2. Treatment intent was assessed by answering yes or no to the question “the goals of my therapy are to get rid of all my cancer,” adapted from Temel et al.¹³

3. Terminal illness acknowledgment was assessed using the question “How would you describe your current health status: relatively healthy but terminally ill, seriously and terminally ill, relatively healthy, or seriously but not terminally ill,” adapted from Coping with Cancer.⁹

Demographic and Clinical Variables

Enrolled patients completed baseline questionnaires providing data on age, sex, race, degree of religious importance, education level, cancer type, and marital status. Patients completed the Hospital Anxiety and Depression Scale (HADS)^44-46 to assess baseline anxiety and depression. ECOG data were collected from oncology assessment as documented in the electronic medical record.

Statistical Analysis

Continuous variables are presented as mean ± standard deviation; categorical variables are presented as frequencies and percentages. Differences in baseline characteristics between the randomly assigned groups are tested using t tests for continuous variables and chi-squared tests for categorical variables. To investigate any potential bias because of analysis of a selected population contingent upon completion of 3-month illness expectation data, we conducted significance tests comparing baseline characteristics of randomly assigned groups to ensure group equivalency.⁴⁷

For each of the illness expectation variables, we calculated the frequency and percentage of participants with realistic illness expectations at baseline and 3 months. We also present the frequency and percentage of patients who remained realistic (realistic response at both baseline and 3 months), became realistic (unrealistic response at baseline but realistic at 3 months), became unrealistic (realistic response at baseline but unrealistic at 3 months) and remained unrealistic (unrealistic response at both baseline and 3 months). We conducted logistic regression analyses to estimate the difference in 3-month illness expectations between groups while adjusting for baseline expectations and selected baseline covariates likely to influence illness expectations (age, sex, race, religious importance, education, marital status, cancer type, ECOG status, depressive symptoms, and anxiety symptoms).^48-51 We suspected that clinic site may also influence patient expectations; on the basis of the cluster-randomized study design, we included a sensitivity analysis using a mixed-effects model with a random effect for site. Finally, we conducted an intensity-adjusted analysis to evaluate the effect of intervention dose on illness expectations by assigning covariates for the number of primary palliative care visits occurring before the 3-month follow-up period. Covariates were defined as 0 for all standard care patients and the number of visits received for CONNECT patients ranging from 0 to 3, then divided by 3 to scale with the regression coefficient providing an estimated effect of full intervention dose at 3 months. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Patient Characteristics

A total of 672 patients were enrolled in the CONNECT trial and eligible for participation, of whom 457 completed the illness expectation questionnaire at baseline and 3 months and were included in this analysis (246 randomly assigned to receive standard oncology care and 211 randomly assigned to receive the CONNECT intervention; Appendix Fig A1, online only). Most participants who did not complete the questionnaire at both time points died before 3 months (n = 143).

CONNECT visits were audio-recorded, 43% of which were evaluated by a nurse project manager to establish and maintain fidelity. Fidelity to assessing views about illness was acceptable, occurring 80.4% of the time during the initial CONNECT visit and 87.3% of the time during subsequent visits.^39,52

Demographic and clinical characteristics of participants are shown in Table 1. The mean age was 69.1 ± 10.2 years. More than half of the patients included in this analysis were female, and 94.5% were White. There were no differences in age, sex, race, marital status, religion, time with current oncologist, or cancer type between groups. At baseline, CONNECT patients were less educated than their standard care counterparts (P = .023). CONNECT patients were also more likely to have a worse performance status (P < .001) and reported higher anxiety symptoms (P = .036) than standard care patients (Table 1).

TABLE 1.

Demographic and Clinical Patient Characteristics

Baseline Illness Expectations

A minority of patients had realistic illness expectations at baseline. No significant differences existed between groups for life expectancy or terminal illness acknowledgment (Table 2). Of 457 patients, 10.0% v 8.9% indicated a life expectancy of months (P = .713) and 46.4% v 42.3% acknowledged being terminally ill (P = .371). Standard care patients were more likely than CONNECT patients at baseline to answer no to treatment intent of get rid of cancer (32.5% v 23.7%; P = .037).

TABLE 2.

Baseline and 3-Month Responses to Illness Expectation Questions

Impact of Primary Palliative Care on Illness Expectations

Both the CONNECT and standard care groups had slightly higher proportions with realistic illness expectations at 3 months compared with baseline (Fig 1). However, there was little evidence of significant differences in realistic illness expectations at 3 months between the CONNECT and standard care groups. Overall, 12.8% of CONNECT patients v 11.4% of standard care patients answered life expectancy of months, 26.1% v 35.8% answered no to treatment intent of get rid of cancer, and 53.6% v 44.7% acknowledged that they were terminally ill (Table 2). These results were little affected when accounting for site variation but were particularly attenuated after adjustment for patient demographic and clinical characteristics (baseline expectations, age, sex, race, religious importance, education level, marital status, cancer type, ECOG status, HADS-Depression, or HADS-Anxiety). Adjustment for treatment dose in the intensity-adjusted analysis did not suggest that a greater dose of the intervention increased the likelihood of realistic illness expectations at 3 months (Table 2).

FIG 1.

Changes in illness expectations by treatment group. Assessments at baseline and 3 months with realistic responses indicated include the following in italics: When you think about how having cancer might affect your life expectancy, do you think in terms of months, years, or do not know? Goals of therapy are no to get rid of all cancer/yes to get rid of all cancer? How would you describe your current health status: relatively healthy but terminally ill, seriously and terminally ill, relatively healthy, or seriously but not terminally ill? There was no statistical difference in the percentage of patients who remained realistic/became realistic between CONNECT and standard care groups at 3 months for life expectancy, treatment intent, or terminal illness acknowledgment. The majority of patients in both groups responded to illness expectation questions with answers that remained unrealistic at 3 months.

DISCUSSION

In this large cluster-randomized trial of an oncology nurse-led primary palliative care intervention, we found little evidence of effect on illness expectations among patients with advanced cancer. Baseline illness expectations were unrealistic in both groups, with most patients responding unrealistically to questions about life expectancy, treatment intent, or terminal illness acknowledgment. Although the primary palliative care intervention did not change illness expectations compared with standard care, there was a nonsignificant trend toward realistic illness expectations in both groups at the 3-month assessment.

Our study builds on a body of work demonstrating that unrealistic illness expectations are challenging to change.^{3,13,43,53,54}

One possible explanation for our negative findings may be that the primary palliative care intervention was delivered by oncology infusion room nurses.⁵⁵ CONNECT nurses assessed patient coping and encouraged discussions with oncologists but did not explicitly address prognosis, which is frequently perceived by registered nurses as outside their scope of practice.⁵⁵ Oncologists were not directly involved in the intervention, which may be significant because the relationship between patients with cancer and their oncologists is unique. The medical training of a provider, the length of time known, and the strength of the relationship may influence patient willingness to trust and apply the content of clinical discussions.²⁷ However, the VOICE trial, which assessed an oncologist communication intervention, failed to affect shared understanding of prognosis of patients with advanced nonhematologic cancer and their oncologists.⁴³ Another possible explanation for our negative findings may be that the intervention was delivered by a member of the oncology team. Patients who were receiving active treatment for advanced disease may have been hesitant to discuss prognosis and life expectancy with their oncology treatment team. In a recent study characterizing communication patterns among oncologists and specialty palliative care physicians, palliative care physicians played a distinct role, tending to assess patient understanding of the treatment process and prognosis more frequently.⁵⁶ It is possible that this additional layer of specialized support for conversations about illness understanding is a key ingredient in fostering realistic expectations.^11,13,18 Future trials may require more involvement from palliative care specialists in discussions to promote illness expectations.

Furthermore, rather than reflecting a purely cognitive understanding of medical facts, we recognize that illness expectations are also shaped by individual medical and nonmedical experiences, beliefs, and coping methods.^1,6,51,57 Fluctuations in illness expectations may reflect changing beliefs and lack of certainty regarding treatment, contributing to the complex and dynamic relationship between emotional responses and cognition.^8,13,52 Furthermore, unrealistic expectations may reflect higher hope. It has been increasingly recognized that false hope may have psychologically therapeutic effects, and it has been associated with increased goal-value congruence, productivity, and well-being in multiple disciplines.^53,54 Oncology treatment advances, such as immunotherapy, have fostered hope among many patients with advanced disease, which may in turn be therapeutic. Despite attending to emotion in the CONNECT intervention, we did not change illness expectations, suggesting that experiences outside the medical encounter may play an important role.

Our findings should be considered in the context of study limitations. This study population included predominantly White patients, which limits the generalizability of our findings to other racial and ethnic patient populations. Prior palliative care studies have recruited participants within 1-2 months of an advanced cancer diagnosis, when illness expectations are forming, coping strategies are developing, and cognitive representations may be more malleable. However, in our study, patients were not recruited within a particular time period after diagnosis. Although we assessed patient illness expectations using questions and outcomes consistent with prior palliative care studies,^12,13,42 these measures have limitations and may be difficult for patients to interpret. For example, 59.7% of CONNECT patients and 54.1% of standard care patients answered unsure to treatment intent questions at 3 months. Validated assessment tools that assess the multidimensional aspects of illness expectations are needed. Oncologists were not directly involved in this intervention, which may reduce oncologist buy-in and limit accessibility of data on specific prognostic discussions between oncologists and patients before or after CONNECT visits. Additionally, only patients who completed baseline and 3-month questionnaires were included in our analysis, and we did not assess illness expectations responses longitudinally after 3 months. Although the intensity-adjusted analysis found no difference by the number of visits provided within 3 months, it is possible that more than three visits over a longer time period may be needed to affect illness expectations. Most patients who failed to complete the follow-up questionnaire died before 3 months. Other patients may have been experiencing transitions in care or declines in health that prevented them from completing the questionnaire; changes in health status and treatment failures may accelerate discussions of prognosis and facilitate more realistic illness expectations, and excluding such data prevents analysis of how such circumstances affect expectations. We acknowledge that this secondary analysis had limited power to detect small treatment effects.

In summary, we observed that an oncology nurse-led primary palliative care intervention showed little evidence supporting an effect on illness expectations in patients with advanced cancer at 3 months. Efforts to improve realistic illness expectations via primary palliative care may require more explicit prognostic discussions and/or earlier integration during the illness trajectory. We also recognize that patient expectations have multifactorial influences, and as the world of cancer treatment continues to evolve, the paradigm of unrealistic expectations and methods for addressing emotional and psychologic needs must reflect these changes.

APPENDIX

FIG A1.

CONSORT diagram.

AUTHOR CONTRIBUTIONS

Conception and design: Lauren E. Sigler, Teresa H. Thomas, Robert M. Arnold, Douglas White, Thomas J. Smith, Margaret Rosenzweig, Kenneth J. Smith, Yael Schenker

Financial support: Yael Schenker

Provision of study materials or patients: Edward Chu, Yael Schenker

Collection and assembly of data: Lauren E. Sigler, Andrew D. Althouse, Douglas White, Yael Schenker

Data analysis and interpretation: Lauren E. Sigler, Andrew D. Althouse, Teresa H. Thomas, Robert M. Arnold, Douglas White, Thomas J. Smith, Edward Chu, Kenneth J. Smith, Yael Schenker

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Effects of an Oncology Nurse-Led, Primary Palliative Care Intervention (CONNECT) on Illness Expectations Among Patients With Advanced Cancer

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).